QFT-0 | Wave Function | Electronvolt

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

2 2. 2. Zuber. Quantum Field Theory. Diagrammatica.V. 4.4 2.3. Veltman.3 3. Peskin and D.3 6.5 4. Peskin & Schroeder and Srednicki. 1.2 4. The notes contain all essential information.8 3.3 4.2 2.4 6. S.5 5.4 2. but are rather compact. M. 2007.1 6.1 4.-B. L. C.2 3. 3. 1995.1 3.5 2. Weinberg. Cambridge University Press.3 2.E. Schroeder.4 3. The quantum theory of fields.1 2. 1994. M.2 3. Vol. Addison-Wesly. Itzykson and J. 1985.4 3. I: Foundations. 1996.2 36 22 .7 2. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012).2 1. Vol.3 3.5 2.H.2 6.1 2 2 3. 5. Quantum Field Theory. These notes Section 1.1 5.1 1.2 3. Cambridge University Press. M.2 3.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. Cambridge University Press. 3. McGraw-Hill.3 2. I refer to the book of Srednicki [1] and Ryder [2].3 References As most directly related books to these notes. Ryder.3 2.3 3. 3.2 6.4 3. Srednicki. Corresponding chapters in books of Ryder. Cambridge University Press.2.1. 6.1 3.3 3. 1980. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. 2.3 2.2 3. II: Modern Applications.3 3. Quantum Field Theory.3 2.2 2. 1995.1 2.4 4. An introduction to Quantum Field Theory. Cambridge University Press.

4.1 9.5 .5 7. 4.2 7. 6.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.4.5. 4.1 10.4 9.3 9.5 6 7 6 6.5 7.2 5.1 20. 6.1 12.2 8.6 9.7.5 2.10 8.1 11.5 4. 2.4 3.2 8. 20. 9.8 5 4.5 4. 4. 6.1 20.5 11.1.4 8.3 8.3 6.2 11. 9. 6.4 2.2 4.3 12.3 7.1.4 4. 4.3 3.3 10.6 3.2.2.1 20.6 6.1 5.3. 2.3 7.7 4.6 8.4 7.1 7.2 10.1.3.5 9.2 12.2.3 2.3 8.1 9.1 8.3 6 8 9 Srednicki 3 3 3 37 5.4 These notes Section 7.4 12.8 6.4 10.7 5.3 11.5 9.6 3.2 9.3.3 20.5 6.4. 6.2.4 12.6 4. 8.6 10.6 3.

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

In quantum mechanics a special role is played by Planck’s constant h. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.6) .Chapter 1 Introduction 1. One can talk about states |r and the wave function ψ(r) = r||ψ . In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-defined concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. In special relativity a special role is played by the velocity of light c. ≡ h/2π = 1. Indeed. usually given divided by 2π. pj ] = i δij . In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r).054 571 68 (18) × 10−34 J s = 6. the action of the momentum operator in the coordinate representation is not as simple as the position operator. Corresponding position and momentum operators do not commute. 1 (1.582 119 15 (56) × 10−22 MeV s. (1. In quantum mechanics. They satisfy the well-known (canonical) operator commutation relations [ri . It is given by pop ψ(r) = −i ∇ψ(r).1 Quantum field theory In quantum field theory the theories of quantum mechanics and special relativity are united. (1. (1.2) In the limit that v ≪ c one reaches the non-relativistic domain. c = 299 792 458 m s−1 . (1.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined.4) where δij is the Kronecker δ function.1) In the limit that the action S is much larger than . (1. S ≫ . the position can in principle be determined at any time with any accuracy. ˜ ψ(p) = i d3 r exp − p · r ψ(r). quantum effects do not play a role anymore and one is in the classical domain. being eigenvalues of the position operators.5) One can also choose a representation in which the momenta of the particles are the dynamical variables.

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.

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

where m is called the mass of the system.27) It is easy to convince oneself of the nature of the indices in the above equation. p). The length squared of ∂ is the d’Alembertian operator.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius.Introduction 6 called covariant. It is an invariant tensor. (1.∇ . ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . constructed via the metric tensor itself.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 .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . (1.33) (1. ∂2 .34) (1. defined by ∂2 − ∇2 .30) (Note that there are different conventions around and sometimes the opposite sign is used). = −24. starting from ǫ0123 = 1. (1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The first identity. for instance. explicitly written as (p0 . g0 = g1 = g2 = g3 = 1. a∞ = 4πǫ0 2 /me e2 . p) = (E. . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . but more important has the same effect on lefthandside and righthandside. .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . because one has (1. 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 . . Note that in this equation one has on left. ′ ′ . Each of these permutations leads to a minus sign.32) . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ fine structure constant α.36) Exercises Exercise 1. (1. ∂3 ) = ∂ ∂ ∂ ∂ . The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. ∂1 . (1. ∂t (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.

0. ǫ0 .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). a+ . a+ . Exercise 1.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). (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). 0.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a2 ] we can find basis vectors n− . 3} the index α can only be equal to one of the indices in ǫµνρσ . n+ . Compare it with the proton mass and use Eq. Also construct and calculate the Planck length Lpl . a2 ] or [a− . 0. ν. but only appropriate combinations. 2. 0. (d) Use the value of the gravitational constant GN / c5 = 6. n1 . This demonstrates how a careful use of units can save a lot of work. g+− and g−+ . n2 . a2 . One does not need to know . ]. Also for the coordinates [a− .13 to give its actual value in kg. a3 ) are the expansion coefficients using the basis vectors n0 . aT ]) are defined through √ a± ≡ (a0 ± a3 )/ 2. a1 . ˆ ˆ ˆ ˆ These are n1 = (0. (b) The coordinates (a0 .3. α and me c2 = 0. ˆ ˆ .2 Prove the identity A × (B × C) = (A · C) B . n2 . by a simple few-line reasoning [For instance: If {µ. re = e2 /4πǫ0 me c2 to the Compton wavelength. How are a+ and a− related to a+ and a− . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. e. 0) ˆ n0 = (1. . 1. c. which is the Compton wavelength for the Planck mass. 0. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . 1. g−− . σ} is a permutation of {0. me . 0.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ .511 MeV. n1 . a1 . [Note: what is the MKS unit for V/T?] Exercise 1. 0) ˆ n2 = (0. . 0) ˆ n3 = (0. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . . a1 . Exercise 1. 1. n3 .(A · B) C using the properties of the tensor ǫijk given in section 1. ρ. 1) ˆ . a+ . 1.

∂t 2M (2. t) = ∂2 − ∇2 + M 2 φ(r. But the replacement 2. t) = − ψ(r. p).3) Equations 2. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive definite.1 and 2. E = p2 .2) Acting on the wave function one finds for a free particle.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. (2. (2. p2 = pµ pµ = E 2 − p2 = M 2 . 8 . (2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. t). t) = 0.Chapter 2 Relativistic wave equations 2. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. Although it is straightforward to find the solutions of this equation.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . 2 + M 2 φ(r.3 are not covariant. namely plane waves characterized by a wave number k. i ∂ ∇2 ψ(r. ∂t2 (2.5) with (k 0 )2 = k2 + M 2 . φk (r. written as pµ −→ i∂µ is covariant (the same in every frame of reference). 2M (2. t) = exp(−i k 0 t + i k · r).7) k 0 = ± k2 + M 2 = ±Ek . ∂t p −→ pop = −i∇.4) where M is the particle mass.1 The Klein-Gordon equation In this chapter. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.2. one obtains (2.

φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . ∂µ j µ = 0. ↔ ↔ (2. which is what Lorentz transformations do. Then. This continuity equation can be written down o covariantly using the components (ρ. 2.e. (2.13) ˜ with (in principle complex) coefficients φ(k). 2M 2M (2. −i φ∗ ∇ φ . It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends.2) is j µ = i φ∗ ∂ µ φ or (ρ. The appropriate current corresponding to the KG equation (see Excercise 2. As we will see later both problems are related and have to do with the existence of particles and antiparticles. And. For this we note that an arbitrary solution for the field φ can always be written as a superposition of plane waves.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. Probability is not positive: in quantum mechanics one has the probability and probability current ρ j = = ψ∗ ψ ↔ i i − (ψ ∗ ∇ψ − (∇ψ ∗ )ψ) ≡ − ψ ∗ ∇ ψ. k) + ei k·x φ(−Ek .9) They satisfy the continuity equation. is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time. rather than as a wave function. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). but rather the zero component of a four vector. it is simply a matter of being careful to find a consistent (covariant) theory.15) . relativistically the density is not a scalar quantity. This leads to the existence of negative densities.11) Therefore. ∂ρ = −∇ · j. 2. (2. j) of the four-current j. having the KG equation as a space-time symmetric (classical) equation. we want the most general solution.14) ˜ ˜ Introducing φ(Ek . j) = ↔ i φ∗ ∂0 φ. −k) . φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2.3).10) ∂t which follows directly from the Schr¨dinger equation. (2π)3 2Ek (2. there are no longer fundamental objections to mix up space and time. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. The KG equation.5). just as the dependence on time was in ordinary quantum mechanics. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.Relativistic wave equations 9 i. for which we need the interpretation of φ itself as an operator.8) (2. however. k) ≡ a(k) and φ(−Ek . there are solutions with negative energy.2 Mode expansion of solutions of the KG equation Before quantizing fields. (2.

changing the sign of the spatial coordinates.Relativistic wave equations 10 In Eqs 2. for which we consider space inversion. Another symmetry is found by complex conjugating the KG equation. This shows how parity transforms k-modes into −k modes. Since a · b = a · ˜ it is easy to see that ˜ b. (2π)3 2Ek (2. 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) .e. (2. The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).17) (2. x ∂µ ∂ µ + M 2 φ(˜) = 0. φP (x) ≡ φ(˜) (P for parity).23) .22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).14 and 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) .19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. after restricting the energies to be positive). (2. Performing some Lorentz transformation x → x′ = Λx.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. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. 2.6 We will explicitly discuss the example of a discrete symmetry.16) The consequence of this is discussed in Exercise 2. x) → (t.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . (2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). 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). (2π)3 2Ek (2. (2π)3 2Ek (2.18) (2. which implies (xµ ) = (t. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). −x) ≡ (˜µ ). i.

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

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

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

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

ˆ ˆ A(ϕ. . in particular that AB → RAB = RA RB ).15) [g.10) σ ˆ · n. n) i ∂ϕ = ϕ=0 (3. H] = 0 for g ∈ G. H] = 0 for g ∈ G. i. n! 2! n=0 ∞ (3. The next issue is to find the appropriate symmetry operators in the Hilbert space. thus. n) ˆ −→ R(2π − ϕ. i. sometimes in more than one way (but this will be discussed later).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. since any finite transformation can be constructed from the infinitesimal ones.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. . . with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. now with radius 2π and with all points at the surface identified (see figure).9) The parameter-space of SU (2). [g. for fixed n.11) Thus σx /2. .e. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. n) π ≤ ϕ ≤ 2π. thus. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. n) −→ R(ϕ. i. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). (3. . . the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. the above mapping corresponds to the trivial mapping of the Lie algebras.e.Groups and their representations 15 where we have used the (for operators new!) definition eA ≡ An 1 = 1 + A + A2 + . (3. immediately sees that the Lie algebras are identical. 2 2 2 (3.12) One.e. Near the identity. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . one has a Lie algebra isomorphism that is linear and preserves the bilinear product. 2 (3. n) ≈ 1 + i ϕ J · n. . For instance in a function space translations. Suppose we have a system described by a Hamiltonian H. SU (2) ≃ SO(3). also can be considered as a filled 3-sphere. 3. The infinitesimal generators of SU (2) are ˆ obtained by considering infinitesimal transformations. 2! . They satisfy σi σj σk = i ǫijk . however. In the full parameter space. (3.14) For a Lie group.

. e. i.e. m = j(j + 1) − m(m ± 1)|j. ˆ U (α. but it is at the heart of being able to construct a suitable Hilbert space. more precise φ(r + a) = U (a)φ(r) where U (a) = exp +ia pop . does in the nontrivial behavior of Poisson brackets of quantities A(x. It may seem trivial. From the algebra one derives J 2 |j. −j. ϕ + α) = U (α. Within the (2j + 1)-dimensional representation space V (j) one can label the eigenstates |j. j − 1. J+ = 0 J− = 0 . just starting off with the (basic) canonical commutation relations [ri . These are mappings from G into the same finite dimensional vector space (its dimension is the dimension of the representation). m . ϕ). θ. Note that the same requirements would apply to classical mechanics. indeed. m = j(j + 1)|j. the commutation relations. m = m|j. The above set of operators H. J− =  2 √           0 0 0 0 2 0 . .6. . p) and B(x. U (τ ) = exp (−iτ H) . j − 1. pop and ℓop work in Hilbert space and their commutators indeed follow the required commutation relations as symmetry operators. . −j one has for j = 0:       Jz = 0 . z ) = exp (+iα ℓz ) . 3. the Poisson bracket operation satisfies all properties of a Lie algebra. m with eigenvalues of Jz . θ. z )φ(r. The latter is sufficient to get the commutation relations of the (quantum) rotation operators. J =  0 0  +  0 0 .Groups and their representations 16 are generated by the derivative acting on the functions.17) (3. In many cases one can simply construct a suitable Hilbert space or part of it by considering the representations Φ of a group G. . To get explicit representations. J+ =  0 1  . ˆ (3.16) with ℓz = −i∂/∂ϕ = −i(x ∂/∂y − y ∂/∂x) or ℓop = r op × pop . ℓj ] = i ǫijk ℓk that are identical to those in Eq.g. Indeed. pj ] = i δij . 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). preserving the group structure. J− =  0      1 0 0  0  . Explicit representations using the basis states |j. In particular the quantum operators ℓop obey in quantum mechanics commutation relations [ℓi . m with m-values running from the heighest to the lowest. These are mappings of G into a finite dimensional vector space. with pop = −i∇. Somehow the nontrivial structure of rotations should show up and it. which preserve the Lie-algebra structure. Jz |j. for j = 1/2:  0  1/2 Jz =   0 −1/2 0 0 −1         .     2  . while J± |j. Similarly one has the time translation operator φ(t + τ ) = U (τ )φ(t). . m . p) (we will come back to this also in Chapter 7). .  0 for j = 1:   1 0   Jz =  0 0   0 0     √  0 0  2 √  0   √  0   0       . . m = j. m ± 1 with 2j + 1 being integer and m = j. for the function space. The finite dimensional vector space just represents new degrees of freedom. Casimir operators commute with all the generators and the eigenvalue of J 2 can be used to label the representation (j). for rotations the operator Jz and the (quadratic Casimir) operator J 2 . it is sufficient to consider the representations Φ of the Lie-algebra G. with H = i ∂/∂t and the rotation operator. one looks among the generators for a maximal commuting set of operators.18) (3. In order to find local representations Φ of a Lie-group G. φ(r. The other generators Jx and Jy (or J± ≡ Jx ± iJy ) then transform between the states in V (j) .

that is any closed curve in the parameter space can be contracted to a point. Consider the j = 1/2 representation of SU(2). the possibility thus exist that some extensions will not be well-defined representations. χ) ≡ e−iϕJz e−iθJy e−iχJz . however.Groups and their representations 17 and for j = 3/2:  0 0  3/2 0   0 1/2  0 0  Jz =   0  0 −1/2 0   0 0 0 −3/2 √   2 √ 0  0       3  . Jz =  i                0 i 0 −i 0 0 0 −i 0 0 0 0 0     . For a decent (welldefined) global representation. For an element in the group G the corresponding point in the parameter space can be reached in two ways. j. .19) (3. 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . 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. SU (2) is the covering group of SO(3). all states in the representation space have the same energy. For SU(2).g. Of all groups of which the Lie algebras are homeomorphic. Given a representation. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2. J− =  2    0        0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0       . χ)|j. θ. m = Dm′ m (ϕ.3). or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . is not simply connected. to be precise the states ǫχ transform via −σ ∗ .    (j) (3. (j) (j) ′ The spin matrices for that Cartesian basis {ǫx . one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ. θ. the conjugate transformation U ∗ acts on χ∗ . For those global representations. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. ǫy . however. one can look at the conjugate representation. J =  0 0    +   0 0   0     0 0 0 0 0 √ 2 0    √   0         . 1 |1. if χ → σχ.20) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. UE (ϕ. i. For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1.e. e. Explicitly. There exist two different types of paths. For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles.      If rotations leave H invariant. This turns out to be the case for all half-integer representations of the Lie algebra. 2 From the (hermitean) representation Φ(g) of G. This works for SU (2). For SO(3). Infinitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. The matrix elements of these unitary representations are known as D-functions. contractable and paths that run from a point at the surface to its antipode. Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states. If a transformation U acts on χ. the conjugate representation is not a new one. 2 |1. dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m . the extension from a local one must be unique. the topological structure of the group is important. m′ |UE (ϕ. the simply connected group is called the (universal) covering group. θ. ǫz } are       0 0 0   0 0 i   0           Jx =  0 0 −i  . Jy =  0 0 0  . −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . however. χ) = eim ϕ dm′ m (θ) e−imχ . If the group is simply connected. The group SO(3). however. 0 ≡ ǫ0 ≡ ǫz .

which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.27) (3. Using (i) and (ii) the Lorentz transformations can be divided into 4 classes (with disconnected parameter spaces) det Λ +1 +1 -1 -1 Λ00 ≥1 ≤ -1 ≥1 ≤ -1 (3.25) (3. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . 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 . det Λ = −1 is called improper). it is convenient to use a vector notation for the points in Minkowski space.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone .22) (Note that xT is a row vector). (ii) |Λ00 | ≥ 1.3 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. σ/2 in a two-dimensional (spin 1/2) case. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. From this property. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. e To derive some of the properties of the Lorentz group. In this section we consider the Poincar´ group. (3. Writing x as a column vector and the metric tensor in matrix form.Groups and their representations 18 3.     0 0   g00 g01 g02 g03   1 0      g     g11 g12 g13   0 −1 0 0  = . 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.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 (3.26) (3.     G =  10 (3. (det Λ = +1 is called proper. e. We considered the generators and looked for representations in finite dimensional spaces.g. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.

This is the origin of the Thomas precession. z ) = exp(i ϕ J 3 ).29)         0 0 1 0         3′    3    V sinh η 0 0 cosh η V i 0 0 0 Note that Lorentz transformations generated from the identity must belong to L↑ . and those of the boosts. (3. [K i . we have found the generators of rotations. z ) ≈ I + i ϕ J 3 . (3. The first two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. 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. [J i . which satisfy the commutation relations (check!) [J i . 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  .   0 0   +1 0    0 +1 0   0   ∈ L↑ (identity)   I =  (3. K j ] = i ǫijk K k . J 3 ). J = (J 1 . z ) = exp(−i ηK 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  . The commutator of two boosts in different directions (e.32)  0 −   0 −1 0      0 0 0 −1 . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. K = (K 1 . Using these explicit transformations. since the generators K do not form a closed algebra.g. it is easy to find the following typical examples from each of the four classes. infinitesimally given by ΛB (η.30)  0 +   0 +1 0      0 0 0 +1   0 0   −1 0    0 +1 0   0   ∈ L↓ (time inversion)   It =  (3. Rotations around the z-axis are given by ΛR (ϕ. one distinguishes rotations 0 + and boosts. infinitesimally given by ˆ ΛR (ϕ. 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). i proof: use ΛT GΛ = G and ΛGΛT = G. J 2 . since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. Returning to the global group. βγ = sinh η. In L↑ . K 3 ).Groups and their representations = i=1 (Λ0 )2 . J j ] = i ǫijk J k .31)  0 −   0 +1 0      0 0 0 +1   0 0   +1 0    0 −1 0   0   ∈ L↑ (space inversion)   Is =  (3. K j ] = −i ǫijk J k . K 2 . z ) ≈ ˆ ˆ I − i η K 3 .28)                 3′    3    V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞.

three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). infinitesimally approximated by (I + ω.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. inf (3.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0.g. 3 30 with η = ω and K = M (e. Of the four parts only L↑ forms a group.39) (e. This is a + normal subgroup and the factor group is the Vierer group.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. ǫ). we have for the (infinitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . We therefore find (again) that there are six generators for the Lorentz group.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . There are six generators.33) These four transformations form the Vierer group (group of Klein). we continue using covariance arguments. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. 3.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 .g. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . which are obtained by writing the infinitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . Summarizing. Also this group can be divided into four parts. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . We will just require Poincar´ invariance for the generators themselves. (3. a). The extension to the Poincar´ group is e ↑ straightforward. (3. K = M ). Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.38) Explicitly. In order to find the commutation relations including the translation generators. P+ etc. [M µν .36) (3. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). J 3 = M 12 ) and for the (infinitesimal) boosts (along (3. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . We now want to find a covariant form of the six generators of the Lorentz transformations. i i0 i i0 1 ijk M jk 2ǫ (3.

J j ] = i ǫijk J k . [J i .48) . The generators of the translations are the (momentum) operators P µ .44) (3. Infinitesimally. . H] = i P i . [K i . (3. that commute e among themselves. P j ] = i δ ij H/c2 . other states in that representation are obtained by the action of operators outside the maximal commuting set. For instance. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . P j ] = [P i . writing the generator P = (H/c.41) 2 At this point. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . P µ |p. [J i . 3. (3. The eigenvalues of these will be the four-momentum pµ of the state.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers.45) Note that the commutation relations among the generators M µν in Eq. [P µ . . . P ) in terms of the Hamiltonian and the three-momentum operators. P ρ ] = 0. H] = [J i . P ν ] [M µν . 3. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. (3. They just state that M µν transforms as a tensor with two Lorentz indices. K j ] = −i ǫijk J k /c2 . check that U1 U2 corresponds with Λ1 Λ2 ). leads to U (Λ. H] = 0. These define states with specified quantum numbers that are eigenvalues of these generators. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ .46) [K i . .42) (to decide on where the inverse is. . To find other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. P j ] = i ǫijk P k . as part of the set. K j ] = i ǫijk K k . .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. a)P µ U † (Λ. 2 2 (3. ǫ) = 1 − ωαβ M αβ + i ǫα P α . [K i . Of the generators of the Poincar´ group we choose first of all the generators P µ .47) (3. = −i (g µρ P ν − g νρ P µ ) . The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the defining four-dimensional representation).38 could have been obtained in the same way. For instance. one obtains 2 [P i . We have here reinstated c. the generators M αβ no longer exclusively act in Minkowski space. Taking any one of the states in an irreducible representation. known from non-relativistic quantum mechanics are obtained. . a) = Λ−1 µ P ν . ν (3. the generators J± in the case of the rotation group. [J i . a)P µ U (Λ. (3.43) giving the following commutation relations by equating the coefficients of ǫµ and ωµν . Explicitly. the generators J 2 and J 3 in the case of the rotation group. a) = Λµν P ν or U † (Λ. These form a group called the little group associated with that four vector. = pµ |p. In order to label the states in an irreducible representation we construct a maximal set of commuting operators.

. Wν ] = −i (gµα Wν − gνα Wµ ) . i ǫµνρσ W ρ P σ . Wα ] = [Wµ .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. 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. . Pν ] = [Wµ . 2 exp (−isµ W µ ) |p. Massive particles: p2 = M 2 > 0. p0 > 0. From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.54) This will enable us to pick a suitable commuting ’spin’ operator. which can thus be chosen spacelike.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . 0. . . 0.52) (3. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3.53) (3.49) (3.55) commute with all generators and therefore are invariants under Poincar´ transformations. the third case represents the vacuum. .56) . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . (3. . for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the first two cases correspond to physical states. . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). = = i exp − ǫµνρσ M µν pρ sσ |p. M ǫijk . . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p.51) The following properties follow from the fact that W µ is a four-vector by construction. These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ . 2 (3. . (3.

e. W j (p)] = i M ǫijk W k (p). W i (p) = −W · ni (p)). −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. Having made a covariant decomposition. More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p). 2 . 3. 1. 0. s.57) (i. M2 (3. 0. |p3 |). 1. 0). M2 (3.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). (1. show that W /M can be interpreted as the intrinsic spin.59) (3. The commutation relations [W i (p).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. p. 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 . . M J) with components W i = (M/2)ǫijk M jk . 0). 0. p0 > 0. In that case the vectors ni (p) are just the space directions and W = (0.61) one sees that S i (p) i =− W2 . (0. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.58) i. 1. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). ms . S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). (3. (0. 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. . .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. it is sufficient to choose a particular frame to investigate the coefficients in Eq. 0.64) such that in an arbitrary frame where the four vectors p. i (3. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. 0. (3. Together with the four momentum states thus can be labeled as |M. 1 2 (3.e.65) . 0).60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 0. n1 (p).Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3..

(3. 3. i W 2 (p). 1 2 24 (3. This state is in physical applications nondegenerate and is denoted by |0 . the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a . a)|0 = |0 . where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2).68) (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 .73) (3. W 2 and λ is derived from the general commutation relations for [Wµ . which can take any integer or half-integer value.1 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. A massless particle. Exercises Exercise 3.Groups and their representations The above does actually include the massive case. W µ (p) = λ(p) P µ . thus. The vacuum: pµ = 0. We don’t know of any physical relevance. W 2 (3.13.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. Note that angular momentum does not contribute to helicity as L · p = 0. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. U (Λ. P µ . (3. 3. W 1 (p)] [λ(p).67) (3. −i W 1 (p). however. The algebra of W 1 . W 2 (p)] = = = i M 2 λ(p). Thus we have P 2 . 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). But if M = 0 one has a different algebra (Eq. is characterized by its momentum and the helicity.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). Thus 2 λ(p) = J ·p . and gives [W 1 (p). σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). |p. λ . implying continuous values for the spin (actually for W i = M S i ).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). W 2 (p)] [λ(p). |p| (3. 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).67 has a vanishing right hand side). Wν ]. 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.71) which is called the helicity.

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

H] + i dt ∂O . ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). sj ] = 0. [ri . momentum and spin operators e satisfy the canonical commutation relations. Do this by showing that the set of operators. [ri .6 Consider rotations and translations in a plane (two-dimensional Euclidean space). Hint: for the Hamiltonian. (a) In classical mechanics. From either of these.69 of the Pauli-Lubanski operators for massless particles. they describe r −→ r′ = r − u t while t′ = t. = r × p + s. r2 ) inevitably leads to interaction terms in the boost operators. p×s 1 (rH + Hr) − t p + . derive the commutation relations and show that the algebra is isomorphic to (i.e. but it can be done. = p. pj ] = [ri . there is a one to one mapping) the algebra in Eqs 3.5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. show first the operator identity [r. sj ] = iǫijk sk . consistent with the (canonical) commutation relations of a quantum theory. if you don’t want to do this in general. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O.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 . Exercise 3. These do not matter in the non-relativistic limit (c → ∞). that’s why many-particle non-relativistic quantum mechanics is ’easy’.Groups and their representations 26 Exercise 3.67 to 3. f (p)] = i ∇p f (p). sj ] = [pi . the others vanish. although that in this case is a relation that is found already from the Ehrenfest relation for O = ri . indeed assures d K /dt = 0. boost invariance leads to a conserved quantity. ..e. (c) Give the unitary operator for boosts. This is possible for a single particle. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. Comment: extending this to more particles is a highly non-trivial procedure. = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. i. in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. pj ] = iδ ij and [si . Exercise 3. rj ] = [pi .e. the operator U (u) that gives U (u) r U −1 (u) = r − u t. H P J K = p 2 c 2 + m2 c4 . Determine the commutation relations for the corresponding operators Ki in quantum mechanics. you might just check relations involving J or K by taking some (relevant) explicit components. i. and U (u) p U −1 (u) = p − m u.

The matrices in SL(2.6) (4.1 The Lorentz group and SL(2. C) is the group of complex 2 × 2 matrices with determinant one. 2 (4. They will be labeled by two angular momenta corresponding to the A and B groups. U (ϕ) H(η) (4. C) Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (hermitean) combinations A = B which satisfy the commutation relations [Ai . B j ] = i ǫijk Ak .2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2).1) (4. K = +i .4) (4. This tells us how to find the representations of the group. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.3) (4. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. 0) Type II : (0. where we restrict (for fixed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. 2 1 (J − i K).8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. j) K = −i J (B = 0). (4. similarly as the homeomorphism between SO(3) and SU (2). becoming the defining representations of SL(2.9) (4. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . respectively. i ǫijk B k .Chapter 4 The Dirac equation 4.5) = 1 (J + i K). Aj ] = [B i . B j ] = [Ai . K = −i . (j. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . It is simply connected and forms the covering group of L↑ . They precisely correspond to the two types of representations that we have seen before.7) From the considerations above. (4. C).10) . 2 2 σ σ J = . 0. 2 2 (4. j ′ ). The group SL(2. With this choice of parameter-spaces the plus and minus signs are actually relevant. K = i J (A = 0). Special cases are the following representations: Type I : (j.

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

The i angular momentum operators J are represented by σ i /2. C). ǫµνρσ will change sign. consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  .1. (vector). (axial vector). The behavior is the same for classical quantities.22) u=     η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. 0) of SL(2.   −→  M (Λ)   (4. the well-known two-component spinor for a spin 1/2 particle. must combine the repe resentations. 4.       = (4. For a particle at rest only angular momentum is important and we can choose ξ(0. + −   1 0    0 1  I2 =    0 0   0 0 0 0 −1 0 0 0 0 1      . the boost in Eq.The Dirac equation 29 where The matrix I2 does not belong to L↑ . m) = χm . m) =            χm uL H(p) 0 . a) and ˜ a = (a0 . and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. We also have seen that the representations (0. A parity transformation changes aµ into aµ . 2 2 (4. i.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. This provides the easiest way of seeing ˜ what is happening. etc.      (4. we have seen that in the rest frame W i (p)/M → J i . 4. but by a transformation belonging ↑ ↑ to the class P− . but rather the operators W i (p) should be used. 2 ) and ( 2 . as we have seen in section 4. It has the same effect as lowering the indices. ) −→ ( .      0   χm   uR   H(p)      . Thus M ∗ and M are not equivalent within SL(2. that have the correct commutation relations. the aij elements of a tensor will not change sign. Taking M (Λ) = H(p). Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). −a). From the definition of the representations (0. 1 ) and ( 1 . 1 ) and ( 1 . 0). For the spin 1/2 representations of the Poincar´ group including parity we.e. (pseudoscalar). they can be connected. but to L↑ . i. −→ −p (vector). e. where a = (a0 . The representing member of the class P− is the parity or space inversion operator Is .g. generators.20) 4. 0) are inequivalent.23) u(p.e. −→ −r (vector). Although these operators cannot be used to label the representations. (scalar). C) are suitable for representing spin 1/2 particles.17. they cannot be connected by a unitary 2 2 transformation.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. m) = η(0. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0.13. Within the Lorentz group. therefore. we obtain for the two components of u which we will refer to as chiral right and chiral left components.

this can be rewritten 1 µ ν {γ .      0 H 2 (p)   uR   uR        . ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. The explicit realization appearing in Eq. We will discuss another explicit realization of this algebra in the next section. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.28) which is a covariant (linear) first order differential equation.33) define for a four vector a the contraction a = aµ γµ / .   (4.  =       (4.27 is known as the Weyl representation.7. which in general is a linear equation in pµ . (4. {γ µ . Applying (iγ µ ∂µ + M ) from the left gives (4.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. γ } ∂µ ∂ν + M 2 ψ(x) = 0.26)        −2    uL uL H (p) 0 or explicitly in the socalled Weyl representation  −M E+σ·p       E−σ·p −M    uR      uL      = 0. Since ∂µ ∂ν is symmetric.29) which is an explicit realization of the (momentum space) Dirac equation. It is of a form that we also played with in Exercise 2.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. The first indeed is solved. The general requirements for the γ matrices are thus easily obtained.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . (iγ µ ∂µ − M ) ψ(x) = 0. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately).32) suppressing on the RHS the identity matrix in Dirac space. (4.31) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p).24) (4. (4. = H −1 (p).The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . From this one obtains the Clifford algebra for the Dirac matrices. 4. (4. 2 (4. We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. γ ν } = 2 g µν . . p (4.

where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.37) Using the explicit form of γ in the Weyl representation (see Eq. one sees that there are two positive and two negative eigenvalues.1: The Dirac sea of negative energy states and antiparticles. 4. the one of the negative energy states. From the vacuum a particle can be removed. remains.27). (4.35) is indeed positive and j 0 can serve as a probability density. This hole forms an antiparticle with the same mass.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. opposite charge).The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. Dirac supposed that the negative energy states were already completely filled and the exclusion principle prevents any more particles to enter the sea of negative energy states. (4.34) for the adjoint spinor ψ ≡ ψ † γ0 . . 4. E = +M (twice) and E = −M (twice). We will see how this is properly implemented (also for bosons!) when quantizing fields. The Dirac sea forms the vacuum. γ =ρ ⊗1= γ = iρ ⊗ σ =    0 −1 −σ k 0 2 We (4.3 General representations of γ matrices and Dirac spinors {γ µ .  .36) (4. This problem was solved by Dirac through the introduction of a negative energy sea. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).g. 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. (4. Relying on the Pauli exclusion principle for spin 1/2 particles. This can most easily be seen in the particle rest frame. The density j 0 = ψγ 0 ψ = ψ † ψ (4. but with properties such that it can be annihilated by the particle (e. γ ν } = 2 g µν .

49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .46) (4. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. 4! (4. 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 .R.3) . γ σ ]. γ 0 ]. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . (4. while L−1 γ µ L = Λµ γ ν .9 and 4. K 3 = M 30 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. For instance in the Weyl representation (but valid generally). 2 (4.  0 (4. −→ Sγµ S −1 . S −1 . (γµ )S. γ 0 = ρ1 ⊗ 1 =   1 0 γµ ψ Different representations can be related to each other by unitary transformations.45) (γ5 )W. = S(γµ )W. γ µ } = 0 and explicitly one has    0 1  1 . We note that ψ → Lψ and ψ → ψ L−1 .  (γ5 )S.47) (4. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.R.43) S=√  .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. that project out chiral right/left states. it is easy to see that PR/L = 1 (1 ± γ5 ) 2    1 0  . γ 2 ].44) and an adjoint spinor defined by ψ = ψ † γ0 .   0 γ k = −iρ2 ⊗ σ k =  k  σ  −σ k  .R.The Dirac equation 32 The Weyl (or chiral) representation:    0 1   .  =ρ ⊗1=  0 −1 3 (4. For example. the rotation and boost generators in Weyl representation in Eqs 4. is   1  1 1    (4.48) are projection operators. −→ Sψ.40) (4.R. which in the case of the Weyl representation are just the upper and lower components. = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . 0 σ3 2 4   1  −iσ 3 0   = − i [γ 3 .41) (4.10 are   1  σ3 0  i   J 3 = M 12 =   = [γ 1 .

59)  −→ ψ c = C ψ =    η ǫ ξ∗ . As with parity we look for a transformation. ∂ with the solution ψ c (x) = ηc Cψ (x). γ5 ] = 0. 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.R. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.57) (4. The latter implies that the conjugate of a right-handed spinor. e. ψR . which is again a solution of the Dirac equation. = =   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.54) e. Explicitly.52)    η ξ one can apply space inversion. x = (t. such that T Cγµ C −1 = −γµ .    −iσ 2   0 −ǫ  = .R.6). Some properties of C are C −1 = C † and C T = −C. =    0 ǫ iσ 2 (4. ∂ (4. This symmetry does not change the spin 1/2 nature. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. In S.    0 −ǫ 0    0   −ǫ 0  .) and all representations connected via a real (up to an overall phase) matrix S. In any representation a matrix C exist. that brings ψ → ψ c . T (4. for instance.g. ψ= (4.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. x) → x = (t. (C)W. T (4. (C)S. is a left-handed spinor.58) where ηc is an arbitrary (unobservable) phase.g. but it does. in Weyl representation we find in Dirac space     T C  ξ   −ǫ η ∗     .53) (4. (i/ − M ) ψ(x) = 0. Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  .R. parity. reverse the charge of the particle.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. usually to be taken unity. (or W.56) (i/ − M ) ψ c (x) = 0.   ψ= (4.R.55) (4. called charge conjugation. One has ψ c = −ψ T C −1 . C is real and one has C −1 = C † = C T = −C c and [C. ∂ (4.

the spin polarization vector in the rest frame is the starting ˆ point. Note that the spinors in the negative energy modes (antiparticles) could be two different spinors. s ) = 0. v(p. 2M 2M (4. with E = Ep = + p2 + M 2 . ¯ ′ ′ Explicit solutions in the standard representation are   χs       .69) = − s In order to project out spin states.The Dirac equation 34 4.67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s′ ) = −2M δss′ . s′ ) = 2M δss′ .1). In that frame is a spacelike unit vector sµ = (0. s).65) (4. s)u(p. s) = u(p. s) + v(k. s) . ψ(x) = u±s (p) e−i p·x . s ) = v † (p.   ⇔ (/ − M ) u(p) = 0.70) =  Ps =  ˆ 0 1∓σ·s 2 2 . s) ei k·x d∗ (k.61)     σ·p −(Ep + M ) There are also two negative energy solutions. s)u(p. s). ¯ (4. s)v(p. s)v(p. s) = v(p..62) where χs are two independent (s = ±) two-component spinors. s) = u(p.g. s)¯(p. s) u = . (2π)3 2Ek (4. where u satisfies   −σ · p  Ep − M      u(p) = 0.64) (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. ¯ ¯ † ′ u (p.     ∂ −iσ · ∇ −i ∂t − M (4. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). In an arbitrary frame one has s · p = 0 and s(p) can e.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions.   ⇔ (/ + M ) v(p) = 0. the spinors satisfy C uT (p. s) e−i k·x b(k. The ¯ ¯ solutions are normalized to u(p. s) v = . the best representation to describe particles at rest is the standard representation. s′ ) = 2Ep δss′ . s) and C v T (p. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       (4.60)   ψ(x) = 0. p (4. It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. s)¯(p. ψ(x) = v ±s (p) ei p·x . where v satisfies   σ·p  −(Ep + M )      v(p) = 0.66) An arbitrary spin 1/2 field can be expanded in the independent solutions. v (p. s ) = v (p. be obtained by a Lorentz transformation.    (4. s)v(p. p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s) = Ep + M      σ ·p   Ep +M χs     . 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. 2M 2M −/ + M p v(p.68) (4. s)u(p.63) u(p.

     (4. This means that in the solution space for massless fermions the chirality states.      u(p. λ) = u(p. We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. In principle massless fermions can be described by twocomponent spinors.74) .       −p3    −p1 − i p2  1   v(p. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). −1/2) = √ (4.72) ˆ Helicity states (p along z ) in Standard representation are:    √  √ 0  E+M           E +M . −1/2) = √  E + M  −(E + M )   0      .48 replace the spin projection operators which are not defined (by lack of a rest frame). λ = − 1 ) = √  √     v(p.71)   p − i p2  . p ψR/L ≡ PR/L ψ are independent solutions. λ = −1/2) =  u(p. For instance with the z-axis as spin quantization axis.           . Note that for solutions of the massless Dirac equation /ψ = 0. while the negative helicity (λ = −1/2) is in essence lefthanded. Explicit examples of spinors are useful to illustrate spin eigenstates. v(p. helicity states. λ = +1/2) =       E−M    √ 0     − E −M 0    √  √ 0   − E−M     − E −M      0  .  2 2   0 √     2 √ 2     − E+M − E−M 0      .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 ).      (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. +1/2) = √ u(p.    v(p.  √ 0   u(p. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ .    1  u(p. 4. one has in standard representation:     0  E+M         E+M        1 1 0   . λ) = v(p. λ = + 1 ) = √     + E +M − E −M  . Therefore. u(p. One finds for the helicity states in Weyl representation: √    √  0√   E +M + E −M   √         1  √ 1  0√ 1  . λ = − 1 ) = √  E + M − E − M  . λ). λ = −1/2) =  √   − E+M     √ 0      0 E+M (4. λ = +1/2) =  v(p. +1/2) = √  0 E +M    E+M      . 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 √  .       p3 E +M  E+M        1 2 3 p + ip −p  1 2  p − ip  3   1 −p   v(p. γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. etc.73) Also for helicity states C uT (p. chirality. The chirality projection operators in Eq. λ) and C uT (p.

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

1 i −i S ρσ = − [γ ρ . γ σ ] = σ ρσ .3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . s)¯(p. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.g. Excercise 4. Show that the equality holds. H(p) = exp η·σ 2 = M +E+σ·p . 0 if n is odd. of which only the traces of four and two gamma-matrices are nonzero. −x) is also a solution of the Dirac equation. e.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4.13. ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . i S ρσ ] = g ρµ γ σ − g σµ γ ρ .g. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. Do this e. Exercise 4. x ˜ . 4. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν .4 Show that in P+ ≡ s=±1/2 u(p.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. by letting them act on the four basis spinors u(p. a 4 a · b.6 Apply space-inversion. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. where x = (t. ˜ Exercise 4.77) Exercises Excercise 4. x → x. 4 2 Exercise 4. s) u p /+M = 2M 2M both sides are projection operators. (4.5 Show starting from the relation {γµ . s). to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). s) and v(p.1 Prove Eq.

   is a stationary solution of the (free) Dirac equation. Note: To prove this. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. 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 ). ν p Note: how such parametrizations work will be explained in chapter 8.511 MeV). |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.e. 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. 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.66 MeV.8 × 10−6 . you need to determine the corresponding confinement condition for ψ. We will not pursue this chiral freedom at this point and take α = 0. M R = 1. A condition for confining the fermion in the cavity is i / ψ = ψ (where n is the surface normal).57 MeV. the decay would even be forbidden. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. Do you understand why? The calculation of the life-time is completed in chapter 11. t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e . there is no current flowing through the surface of the cavity. which is the most important factor needed to find the decay width (inverse lifetime). (b) Next. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. . i. Excercise 4. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. n (c) Apply the confinement condition to find the value of k for the lowest eigenmode in the spherical ˆ cavity. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. For such a cavity nµ = (0. mµ = 105.5 and M → ∞. we want to confine this solution to a spherical cavity with radius R. Show that this n condition is sufficient to guarantee that nµ j µ = 0. namely i / ψ = eiαγ5 ψ.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). r). Calculate for now just |M |2 . The pion has spin 0.The Dirac equation 38 Exercise 4. n It turns out that there is actually a whole family of conditions that provide confinement.7 (a) Show that (in standard representation)  j0 (kr)χm    ψ(x. me = 0. Calculate the value of k for M R = 0.

0). one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . A real vector field Vµ (x) then can be expanded in modes as Vµ (x) = λ=0.± 5. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . These can describe spin 1 states in the rest-frame. ±(1) are obtained by boosting the restframe vectors. (2 + M 2 )Vµ = 0. M2 (5. The vectorfield is therefore (λ) suited to describe spin 1.4) λ=0.1) The Lorentz condition implies that k µ ǫµ (k) = 0. µ (2π)3 2E µ (5. there are then indeed three independent possibilities.Chapter 5 Vector fields and Maxwell equations 5.  −B 1    0 3 With the electromagnetic field tensor (5. In an arbitrary frame ǫµ (k) with λ = 0. λ) .2) (5. with ∂µ V µ = 0 (Lorentz condition). λ) + ǫ(λ)∗ (k) eik·x c∗ (k. ǫλ ).2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. namely ǫµ (k0 ) = (0.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.6) where Aµ is the four-vector potential (5. where k = k0 = (M. These are solutions of the Klein-Gordon equation with in addition a condition.1 Fields for spin 1 In section 3.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. In the restframe.7) . just as the two-component Dirac spinors did for spin 1/2.2 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. A). 39 (5.5)  −E    B2   . (5.

˜ x Note that via parity transformation one obtains another solution. For electrodynamics one has furthermore the freedom of gauge transformations.17) (radiation. AP (x) = Aµ (˜). 0. One possible choice is the gauge A0 = 0 (5.Vector fields and Maxwell equations and the (conserved) current j µ = (ρ. if in Eq. As remarked above.5. ∂µ Aµ = 0. The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). (5.15) (5. 2Aµ = 0.14) of which the solutions give the Li´nard-Wiechert potentials. (5. 0). The equation in vacuum. it will vanish at all times. in accordance with what we found in section 3.12) (5.18) corresponding to the helicity states λ = ±1 of the massless photon field.13) Hence.11) This equation is not affected by a gauge transformation. −i.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. Although the Lorenz condition is a constraint.8) (5. 5. in which case one has 2Aµ = jµ . Therefore if k µ = (|k|. 40 (5. Taking the divergence of this equiation. (5. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. ∓1. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. (5.15 the d’Alembertian is replaced by 2 + M 2 . The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . transverse or Coulomb gauge). j) the Maxwell equations read ∂µ F µν = j ν . Furthermore. shows that the electromagnetic fields correspond to the case of a massless spin 1 field. only two independent vectors remain ǫ(±) µ (k) = (0.16) (5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field.10) the electric and magnetic fields are unchanged. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field. |k|). 0. moree over. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. for M = 0. if ∂µ Aµ = 0 for large times |t|. i. It states the absence of magnetic charges and Faraday’s law. one finds for a conserved current and λ = 0 2∂µ Aµ = 0.e. (5. .

ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. .0 . cos ϕ. It is however. 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 . but where E = B = 0. λ L − λ λ or ∆x = (L− λ/d)δ. In this experiment an observable phase difference is measured for electrons moving through field-free space (E = B = 0 but Aµ = 0). Inside solenoid: Outside solenoid: B = B z. 5. This causes an interference pattern as a function of ∆x. F = e E + ev × B (5. . 2 2 2 BR2 BR2 y BR2 x A= .1.Vector fields and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5.0 . This phase difference is actually observed. while the electromagnetic field Aµ is not. (5.1: The Aharonov-Bohm experiment 5. A= By Bx Br ϕ= − ˆ .20) e (∇ × A) · ds = B · ds = where Φ is the flux through the solenoid. In the presence of an electromagnetic field Aµ an additional phase difference is observed. sin ϕ. . nevertheless. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum).19) are gauge invariant. This occurs in the region outside the solenoid. but A = 0 (hence there is structure in the vacuum). 0). may look a bit akward. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). illustrated in fig. 0) and ϕ = (− sin ϕ. The situation.3 The electromagnetic field and topology∗ The electric and magnetic fields E and B as appearing for instance in the Lorentz force on a moving charge. The phase difference between the electrons travelling two different paths is given by δ = 2π a a ∆x d = −≈ . = exp −i The phase difference between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. To be precise. Nevertheless the significance of Aµ is shown in the Aharonov-Bohm experiment. where Aµ = 0. ˆ B = 0.

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

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

and the total change x′ (t′ ) = x(t) + ∆x(t). x) = K − V = ˙ 1 mx2 − V (x). (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. ∂x ˙ (6.Chapter 6 Classical lagrangian field theory 6. thus t′ = t + δτ. t1 (6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation).3) (6. These equations can be obtained from a lagrangian using the action principle. ˙ p= ∂L . As an example. recall classical mechanics with the action t2 S= t1 dt L(x. x (t) = x(t) + δx(t). ∂x (6. x) ˙ (6. d dt ∂L ∂x ˙ = ∂L .4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 .1) and as an example the lagrangian L(x. ˙ 2 (6.6) The first term leads to the Euler-Lagrange equations.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current.8) 44 .5) with ∆x(t) = δx(t) + x(t) δτ . The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.

indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ .17) Furthermore the variations δφ and δ∂µ φ contribute to δS. ∂µ φ(x)) = R d4 x L (φ(x).t1 ) (6. . think for instance of a temperature or density distribution or of an electromagnetic field).18) 1 Taking a functional derivative. t1 ) and R3 . i. ∂µ φ. with ∆φ(x) = δφ(x) + (∂µ φ)δx . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). their derivatives and possibly on the position. ∂R Note for the specific choice of the surface for constant times t1 and t2 . ∂µ φ(x)). but which in 3 dimensions includes conserved quantities related to rotations and boosts. x′ ) − d4 x L (φ.Classical lagrangian field theory 45 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . . Consider a lagrangian density L which depends on these functions. . t1 (6.15) d4 x′ L (φ′ . The hamiltonian H is defined by H(p. R (6. . dσµ .2.14) (6. (R3 . x) ≡ p x − L .11) λ=0 of which p and H are the simplest examples related to space and time translations. .6 can be rewritten as δS = . 6. = φ(x) + δφ(x) (6. Variations in the action can come from the coordinates or the fields. λ=0 (6. − d3 x .12) Here R indicates a space-time volume bounded by (R3 . δ(∂µ φ) (6.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . In general any continuous transformation. . x). . . ˙ x The second term in Eq.t2 ) d3 x . t2 ). ∂µ φ′ .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ).13) (6. L (φ(x). ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . indicated with δF [φ]/δφ should pose no problems. H is a conserved quantity. We will come back to it in a bit more formal way in section 9. which in classical mechanics vanishes at the boundary) does not play a role.e. ∂µ φ(x). In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields. + ∂L ∆x − H δτ ∂x ˙ t2 . . = ∂R (R3 . (6. .9) which is done because the first term (multiplying ∆x. ˙ (6. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x).

The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the lagrangian densities. For the complex scalar field one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ.27) (6. leading to (2 + M 2 )φ(x) = 0. ∂ (6. (6. The integrand of the first term must vanish. 2 ∗ (6.25) (6.19) δ(∂µ φ) δφ 6.Classical lagrangian field theory 46 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. (6. δL δL ∂µ = .2 Lagrangians for spin 0. the second term is irrelevant.26) The Dirac field The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ.22) which can be considered as the sum of the lagrangian densities for two real scalar fields φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. One easily obtains (2 + M 2 )φ(x) = 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). δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. Using the variations in ψ (in the symmetric form). the Dirac field (spin 1/2) and the vector field (spin 1) can be found.24) (6. leading to the Euler-Lagrange equations.29) .23) (6.21) which differ only by surface terms.20) (6. 1/2 and 1 fields By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar field (spin 0). 2 (6. 2 ψ = 0. (2 + M )φ (x) = 0. (6.

We note that of which the left part coincides   . Peshkin and Schroeder or Exercise 12.31) showing e. or equivalently separate the field into right. and thus (β ∗ α)∗ = α∗ β = −βα∗ .34)  (6.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . but note the factor 1/2 as compared to the Dirac lagrangian.36) (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. . L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .and lefthanded ones.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. 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.31 is given by LM (Dirac) = −M ξ † η + η † ξ . Υi / Υ − ∂ 4 2   T  −ǫ η ∗  c  = ΥR .g. 1 M η † ǫη ∗ − M ∗ η T ǫη .32) There exists another possibility to write down a mass term with only one kind of fields. The reasons for Grassmann variables will become clear in the next chapter. (6. The Majorana case is in fact more general.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). 2 αβ = −βα. namely2 LM (Majorana) = + (6. ∂ ∂ 2 2 (6.g.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). the mass term in the Dirac lagrangian 6.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. (6.    ∗  0    ⇒ Υ ≡  −ǫ η  ΥL = ψL =    η η ∗ for which η0 = ǫ η0 . 2 (6. Using the twospinors ξ and η.38) (6. that the lagrangian separates into two independent parts for M = 0.Classical lagrangian field theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.7). 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. which comes because we in essence use ’real’ spinors. ψL ≡ (ψL )c = C ψL =    0 (6.

42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . requires the presence of a conserved quantity. ∂µ J µ (x) = 0. (6. .41) (6.46) 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 . These do not affect the dynamics and will give rise to conserved currents. δ(∂µ φ) (6. t). In particular when the lagrangian is invariant under symmetries.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). t). ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. which in a consistent picture precisely generate the symmetries.18. 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. (6. (6. discussed in the next chapter.. one can consider the variations at the initial or final surface. (6. we consider the second term in Eq. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x).3 Symmetries and conserved (Noether) currents Next. In this case. As an example for the classical field. + ∂R dσµ dσµ ∂R ≡ where d x . (6.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . δ(∂µ φ) (6. Returning to δS the surface term is rewritten to δS = R d4 x . the presence of a symmetry that leaves the lagrangian invariant.44) δL ∆φ + Θµν (x)δxν .47) . consider U (1) transformations of the Dirac field proportional to the charge e.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . . these conserved quantities become operators.σ1 with σ = (R3 .Classical lagrangian field theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . δ(∂µ φ) For continuous transformations. 6. Q(t1 ) = Q(t2 ).. Q= d3 x J 0 (x. which is the space-integral over the zero-component of a conserved current.40) 6.

e.Classical lagrangian field theory From the lagrangian for the Dirac field.49) These currents are conserved as discussed already in chapter 2.48) For the complex scalar field the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .g. (6. 6. Summarizing. For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields. including e the space-time translations and the Lorentz transformations.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. H = d3 x Θ00 (x). ↔ (6. (6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). t) is the conserved charge (Q = 0).4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. In the next section we will see the quantity show up as the charge operator.54) (6.55) Note that the coordinate transformations can be written in a form similar to that for the fields (matrices in internal space).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). Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν . we obtain (since δxµ = 0).50) (6. (6. = 0. omitting the parameter Λ jµ = − δL δL (i e ψi ) (−i e ψi ) − δ(∂µ ψi ) δ(∂µ ψ i ) 49 = e ψγ µ ψ. The integral over the zero-component. 1 (6. ˙ Q = d3 x j 0 (x. (6.53) These are the energy and momentum. From Noether’s theorem one sees that the current Θµν ǫν is conserved.57) For the fields the behavior under Lorentz transformations has been the subject of previous chapters.51) (6.3) j µ = i e φ∗ ∂µ φ. 2 (6. .

58) (6.59) (6. In general this tensor is not symmetric.63) µν = ∂µ Θ . is not invariant under local gauge transformations. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ . in which the U (1) transformation involves an angle e Λ. (6. (6. ∂µ φ(x) + i e ∂µ Λ(x) e (6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.3 we have seen global transformations or gauge transformations of the first kind. Defining T µν = Θµν − ∂ρ Gρµν . where H ρµν is the quantity appearing in M ρµν .g. • Dirac field ψ: −i Sρσ = −(1/4)[γρ . γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4.Classical lagrangian field theory 50 • Scalar field φ: −i Sρσ = 0.3).67) i. 2 (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 . Any lagrangian containing derivatives. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. 6. independent of x. φ(x). • Vector field Aµ : −i Sρσ = aρσ . φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).64) (6.e.6) T µν = T νµ . ∂µ T µν (6.65) M µρσ = T µρ xσ − T µσ xρ . In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). (6. however. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).68) (6. −i eΛ(x) ∗ i eΛ(x) (6. one has a tensor that satisfies (exercise 6.70) .69) i eΛ(x) φ (x). (6. (6.66) In section 6.61) A final note concernes the symmetry properties of Θµν . e. the angle of the transformation depends on the space-time point x.

(6. D ∂ A (6.72) (6. (6.Classical lagrangian field theory 51 where the last term spoils gauge invariance.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 .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). 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).75) The field φ is used here in a general sense standing for any possible field. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). ∂µ φ) =⇒ L (φ. ∂ 2 4 (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic field. 4 (6. Dµ φ) − Fµν F µν .71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. 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.73) (6. −e j µ Aµ = −e ρφ + e j · A. (6. 1 L (φ. 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.76) We note first of all that the coupling of the Dirac field (electron) to the vector field (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ . For this purpose it is necessary to introduce a vector field Aµ . 2 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 . As an example consider the Dirac lagrangian.78) . i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.

i. and deduce what is the charge of the antiparticle as compared to a particle. −eψγ ν ψ. Exercise 6.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar field is connected to a U (1) transformation on the fields.Classical lagrangian field theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . 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.e. ∂µ F µν = j ν . (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. (6.1 (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 .79) Exercises Exercise 6. φ → eiΛ φ. 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. . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. ↔ Exercise 6. where j µ = e ψγ µ ψ. ∂µ F µν = j ν . give the equation which is satisfied by ψ c = Cψ . giving the Maxwell equation coupling to the electromagnetic current. Exercise 6.4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T field. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion).2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .

It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic field takes the value gs = 2. Hint ≡ −gs µe · B = −gs Qe s · B. 2M 2M with Enr ≡ E − M ≪ M . finally note that it is sufficient to show the last equation for dσµ M µρσ . Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices.4) for a stationary solution in the non-relativistic (|Enr | ≪ M ) and weak-field (|A0 | ≪ M ) limits yields in the standard representation for the ’upper (two) components’ ψu .4. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. . Comment: this relation is known as the Pauli equation. use ∂µ T µν = 0 and ∂µ M µρσ = 0. t). This relation is known as the Pauli equation.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . A) (see also exercise 6.6 (optional) Prove the properties of the tensor T µν in section 6. t) = Enr ψu (r.Classical lagrangian field theory 53 Exercise 6. in order to show that Θρσ − Θσρ = ∂µ H µρσ . 2me Exercise 6.

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

14) In fact. ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ). ∇φ) (7. t)] = 0.8) (7. t)] = i δ 3 (x − y). The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). • Poincar´ invariance e The full variation of the fields. the discretization procedure above is an explicit example. formally. py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x. φ.16) . we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. discussed in the section 6. ∆φ. (7.2. qx+∆x ) . t).15) with furthermore the relations [φ(x. 2 2 2 (7. (7. ∂ qx ˙ ∂ qx ˙ (7. ˙ δ φ(x) (7. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta. φ(y. Π(y. t). The essential conditions to consistently quantize a theory then are the following.11) (7. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . t).Quantization of fields 55 etc. qx . px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).7) d3 x L (φ(x).9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν .6) ∆3 x L x (qx . for symmetry transformations. (7. As an example. Φ(f ) = d4 x φ(x) f (x). Π(y. Sometimes it is useful to keep in mind that. for the scalar field theory. a) φi (x)U (Λ.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. i U † (Λ.12) (7. This implies specific transformation properties in the Hilbert space for the quantum fields φ. the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . t)] = [Π(x. ∂µ φ(x)) = ˙ d3 x L (φ. albeit with ’sharp’ functions.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum field theory. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). a) = Rj (Λ−1 ) φj (Λ−1 x − a).

H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ .4.24) It is straightforward to find the commutation relations between N and a and a† . (7.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. (7.17) (Sµν )i φj (x). P ] = i. a] = [a† . [N. [φ(x).23) The hamiltonian in this case can be expressed in the number operator N = a† a.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. 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.Quantization of fields 56 or if U (Λ. j 2 one has [φi (x). a† ] = 1. sometimes referred to as second quantization. j (7.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. P ] = 0 (7. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. 2 i ωµν (S µν )i .18) with S µν given in 6. The measurements cannot influence each other or [Φ(f ).19) 7. a† ] = 0.2 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. [a. Φ(g)] = 0. In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q .25) . a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . φ(y)] = 0 if (x − y)2 < 0. Microscopic causality implies local commutativity. a† ] = a† . and [N. [Q. (7. Q] = [P. Mµν ] = i i∂µ φi (x). i(xµ ∂ν − xν ∂µ )φ (x) + i (7. ω a† a + 2 2 ω i − [Q. a] = −a. 2 2 (7. • 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. Pµ ] = [φ (x).

a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). 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).Quantization of fields 57 Defining states |n as eigenstates of N with eigenvalue n. a† (k ′ )] = 0.33) . † √ a and a act as raising and lowering operators. which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them.g. i. 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 . [φ(x). 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. and we see that a state |0 must exist for which N |0 = a|0 = 0. Π(x )]x0 =x′0 = 0. φ(x) = (2π)3 2Ek where the Fourier coefficients a(k) and a† (k) are now operators. 7. e. the real scalar field φ(x) becomes d3 k (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . but one needs to realize that in that case k 0 = Ek = k2 + M 2 .29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .28) (7. φ(x )]x0 =x′0 = [Π(x).4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).e.3 The real scalar field We have expanded the (classical) field in plane wave solutions.26) a(k) e−i k·x + a† (k) ei k·x .30) [φ(x). (n − 1) a|n † i. (7.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. and (7. The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion. is equivalent with [a(k).e.like’ commutation relations between a(k) and a† (k ′ ). N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n . which represents the ’number of particles’ with momentum k. Note that we will often write a(k) or a† (k). (2π)3 2Ek (7. a(k ′ )] = [a† (k).32) (7.

The rest of the states are then obtained from the groundstate via the creation operator.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). k 0 = Ek = k2 + M 2 ). i. .e. . writing all annihilation operators to the right of the creation operators. (2π)3 2Ek (7. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . 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 (7.39) The problem with the vacuum or zero-point energy.35) where the vacuum contribution disappears because of rotational symmetry. n1 ! n2 ! (7. d3 k a(k) e−i k·x . which amounts to redefining H as H= d3 x : H (x) := d3 k Ek N (k). is solved by subtracting it as an (infinite) constant. d3 k |k k|.46) (7.41) (7.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. in field theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek .45) (7.42) (7. . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). |k = a† (k)|0 . (2π)3 (7. defining particle states |k = |k (with positive energy. (2π)3 2Ek (7.43) 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). 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 (7.37) (7.40) This procedure is known as normal ordering. a(k)|0 = 0. and multiparticle states |(k1 )n1 (k2 )n2 . which now contains an infinite number of oscillators. assuring that the vacuum (by definition) has eigenvalue 0!. This term will be adressed below. For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts.44) p|φ− (x)|0 = ei p·x . (2π)3 2Ek d3 k a† (k) ei k·x .34) where V = (2π)3 δ 3 (0) is the space-volume.Quantization of fields 58 where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy. . For the momentum operator one has Pi = d3 x Θ0i (x) = 1 2 d3 x ∂ {0 φ ∂ i} φ = d3 k k i N (k). |0 . (2π)3 2Ek (7. .

(2π)3 2Ek ≡ i∆− (x − y). φ(y)] consider µν [φ+ (x). The last item to be checked for the scalar field are the causality condition.52) The result for the invariant commutator function is [φ(x).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.18).51) (7. = − (2π)3 2Ek (7. φ− (y)] ≡ i∆+ (x − y).50) or as integrals over d4 k.53) t=0 .17) and (7. d3 k ei k·(x−y) = −i∆+ (y − x). In order to calculate [φ(x).56) = −δ 3 (x).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. (7. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . d3 k = (2π)3 2Ek = (7. 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 ). (7. 2π (7. (2π)4 (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. φ+ (y)] d3 k e−i k·(x−y) .e.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .47) ′ d3 k ′ e−i k·x+i k ·y [a(k). that they satisfy the required commutation relations in Eqs (7. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. (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. ∆(x) = − ǫ(x0 ) δ(x2 ).49) (7. (iii) ∆(0. i.48) (7.

2 From the lagrangian density (7.62) (7. extending it with the charge operator and the introduction of particle and antiparticle operators.e. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.63) i. (2π)3 2Ek (7. 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) . b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . = ∂{µ φ∗ ∂ν} φ − L gµν . (7. a† (k ′ )] = [b(k).64) The commutator of φ and φ† is as for the real field given by [φ(x).60) = i φ∗ ∂µ φ. ↔ (7.4 The complex scalar field In spite of the similarity with the case of the real field.61) (7. φ† (y)] = i∆(x − y).58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). (2π)3 2Ek (7. Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) . we will consider it as a repetition of the quantization procedure.66) . ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).Quantization of fields 60 7.65) 7.5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ. which is equivalent to the relations (only nonzero ones) [a(k).59) (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).57) (7. The field satisfies the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized fields are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (2π)3 2Ek (7.

74) (7. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.71) (7. s)¯(k.77) Note that achieving normal ordering. (2π)3 2Ek (7.70) where the last line is obtained by using the Dirac equation. {b(k. s)u† (k. s)e−i k·x . (7.67) i ψ / ψ − M ψψ gµν .79) Also for the field and the canonical conjugate momentum anticommutation relations are considered.78) (7. s)e−i k·x + d† (k. b† (k ′ . The quantized fields are written ψ(x) = s d3 k b(k. s)¯(k. ↔ i ψγµ ∂ν ψ − 2 (7. s). ∂ 2 ↔ (7.Quantization of fields 61 the conserved density and energy-momentum currents are easily obtained. s) e−i k·(x−y) j s † + vi (k.73) (7.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). s). s′ )} = {d(k. interchanging creation and annihilation operators. The solution is the introduction of anticommutation relations. u v (2π)3 2Ek (7.75) (7. jµ Θµν = = ψγµ ψ. (2π)3 2Ek d3 k b† (k. . s)ei k·x . s) ei k·(x−y) x0 =y 0 .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) : .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. (7. i. (2π)3 2Ek (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . 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) . † {ψi (x). d† (k ′ . ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. s)vj (k.e.69) (7. s)v(k. s)u(k. s)ei k·x + d(k.

−λ(∂ρ Aρ ). It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . 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. For the scalar combination {ψi (x).82) where i∆1 = i∆+ −i∆− . (7. but has the problem of being noncovariant. ˙ δ A0 δL = F i0 = E i . one has † {ψi (x). we would have obtained [ψi (x).6 The electromagnetic field 1 L = − Fµν F µν . When we would have started with commutation relations for the field ψ and the canonical momentum.80) 3 = δ 3 (x − y) δij .84) (7. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i.83) From the lagrangian density the canonical momenta are = = (7. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. as the vanishing of Π0 induces a constraint.Quantization of fields 62 Using the positive and negative energy projection operators discussed in section 4.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. 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). 4 Π0 Πi δL = 0. ∂ (7. ψ j (y)} at arbitrary times one has {ψi (x).88) . (7. ψ j (y)] = (i/x + M )ij i∆1 (x − y). 4 2 This gives the equations of motion discussed before. ˙ δ Ai (7.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before.85) which reflects the gauge freedom.86) (7. ∂ (7. ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y).

0)c(k. λ)ei k·x . These problems are solved by the Lorentz constraint between physical states given above. [ℓi .e. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. j = 1. 0) . (d) Use Poisson brackets to calculate q and p. where Q is the conserved quantity in Eq. (b) Proof for a quantity O(p. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). Aν (y)] = [Πµ (x). what are [ri . 0. 7. λ)|A = 0. a longitudinal photon and two transverse photons. containing a time-like photon. and are equivalent with [c(k. The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. where Aµ + + is the part of the vector field containing the annihilation operators.91) (7.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). µ λ=0 (7. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.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.Quantization of fields 63 for physical states |A and |B . λ) − c† (k. In fact the commutation relations imply ˙ [Aµ (x). 2. It gives 3 k µ ǫ(λ) (k)c(k. 0.94) Choosing k µ = (|k 3 |. The canonical equal time commutation relations are [Aµ (x).89) with four independent vectors ǫµ . ˙ ˙ . rj ]P . c† (k ′ . ′ (7. i. Q]P . which reproduce the Hamilton equations.3) satisfy the conditions of a Lie algebra. (7.e. µ λ=0 (7. λ)e−i k·x + c† (k. i. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). pj ]P . 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.11. q) that dO/dλ = [O. pj ]P . (7.95) Exercises Exercise 7. λ). for which it is sufficient that ∂µ Aµ |A = 0. (7. Πν (y)]x0 =y0 = i gµν δ 3 (x − y).1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. and [ℓi . 6.93) which does exhibit problems with the time-like photon. Πν (y)] = 0. λ)c(k.

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

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

starting with the Dirac equation for ψ(x). writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). parity (P). Table 8. time reversal (T) and charge conjugation (C).1 Parity xµ = (t. x (8. and C quantity t r xµ E p pµ L s ˆ λ = s·p P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ T -t r −˜µ x E -p pµ ˜ -L -s λ C t r xµ E p pµ L s λ 66 . T and C for classical quantities is shown in the table 1. −r).2) where ηP is the intrinsic parity of the field and A is a 4 × 4 matrix acting in the spinor space.1) The parity operator transforms We will consider the transformation properties for a fermion field ψ(x). 8.1: The behavior of classical quantities under P. We can determine A. The consequences of P. r) −→ xµ ≡ xµ = (t.Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. (iγ µ ∂µ − M ) ψ(x) = 0. ˜ (8. T. Both ψ p and ψ satisfy the Dirac equation.

7) = λ d k x x ηP b(k.4) (8. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). x The latter behavior of the vector field will be discussed further below. For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). −λ)u(k.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. m) = = ˜ u(k. x (8. λ)γ0 u(k. 8.6) (check this for the standard representation. (i˜ µ ∂µ − M ) ψ(˜) = 0.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k.5) (8. λ) e−ik·x − d† (k.e. −λ) e−ik·x − d† (k.9) (8. −λ) e−ik·x − d† (k. (2π)3 2Ek (8. In the same way as the Fermion field. the above operation reverses the sign of λ). −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.11) (8. if helicity λ is used instead of the z-component of the spin m. . λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. λ) P −1 = −η ∗ d(k. λ)u(k. λ)γ0 v(k. γ x (8.10) i. λ)v(k. m). λ) eik·x . −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. −λ). −λ). x It is straightforward to apply this to the explicit field operator ψ(x) using γ0 u(k.3) (8. ˜ −v(k. ˜ Pop d(k. λ) e−ik·˜ + d† (k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ)u(k. −λ)v(k. op P (8. m). iγ µ† ∂µ − M ψ(˜) = 0. λ)v(k.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. one can also consider the scalar field and vector fields.12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). m) γ0 v(k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). λ) Pop = ηP b(k.

17) 3 = λ This shows that −1 Cop b(k. Top |φ = φt | = (|φt )∗ . λ) e−ik·x + b† (k.15) (where one must be aware of the choice of spinors made in the expansion. x iγ µT ∂µ − M ψ(−˜) = 0. we start with the complex conjugated x Dirac equation for ψ. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). .e. r) −→ −˜µ ≡ −xµ = (−t. x (8. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. λ)C uT (k. u(k. m). (8. m) ¯ = = v(k. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. x x (8. λ). m). antilinear operator is anti-unitary if A† = A−1 . The same relations hold for helicity states. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. λ) −1 Cop = ∗ ηC b(k. it is antilinear1 . as discussed in section 4). λ). (8.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0.18) (8. to find ψ t (−˜) that is a solution of the Dirac equation. λ)u(k. λ)v(k. x (8.20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). The (time-reversed) Dirac equation becomes.3 Time reversal xµ = (t.21) where A is a 4 × 4 matrix acting in the spinor space. λ) Cop = ηC d(k. For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.Discrete symmetries 68 the latter being also a solution of the Dirac equation. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. (2π)3 2Ek (8. As time reversal will transform ’bra’ into ’ket’. r). λ)C v T (k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. λ) e−ik·x + b† (k. λ) eik·x . One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ .14) (8. Norm conservation requires Top to be anti-unitary2 .16) d k ηC d(k. i. m) ¯ C v T (k. 8.19) Cop d(k.

˜ Top d(k. λ). λ)u∗ (k. λ| ¯ C |¯. λ a |a.32) (8. −p. −λ |¯.26) d k x x ηT b(k.24) (8. a state |a.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit field operator ψ(x) using i γ5 C u(k. λ |¯. λ)iγ5 Cv(k. p. (8.29) In table 2 the behavior of particle states under the various transformations has been summarized. op T (8. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ)v ∗ (k. λ) T −1 = η ∗ d(k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . −λ a T a. x (8. λ) e−ik·x (2π)3 2Ek (8. λ)v ∗ (k.34) . λ)u∗ (k.31) (8. λ)v ∗ (k. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. p. λ a P |a. −p. −p. λ) i γ5 C v(k. ˜ = 0|b(k)|A 8. λ). λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). λ)iγ5 Cu(k. λ). ∗ ˜ v (k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. p. Since there are 16 independent 4 × 4 matrices. there are 16 independent of these bilinear combinations. 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) eik·x + d† (k.25) (check this for the standard representation). λ). λ) eik·x + d† (k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ)u∗ (k. −p. The result is = = λ (8. λ) Top = ηT b(k.30) (8. λ) eik·˜ + d† (k.33) (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ) eik·x + d† (k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k.28) (8. p. λ| a. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k.Discrete symmetries 69 Table 8.

35) The x-dependence can be accounted for straightforwardly using translation invariance. p)u(p) e−i (p−p )·x . p′ . s′ |V µ (x)|p. e.40) where the coefficients are functions of invariants constructed out of the momenta. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .41) 2M 2M . (q 2 ). γ ν ]. 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 ). γ5 γ µ or γ5 ). for the vector current V µ (x).. 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 . ¯ ′ ′ (8.Discrete symmetries 70 Table 8. F. For instance for nucleons one has Γµ (p′ . as well as under the combined operation Θ = PCT (see Table 3). s . ¯ In general the expectation value between momentum states can be more complicated. in this case only q 2 = (p′ − p)2 . p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). p′ |V µ (x)|p = u(p′ )Γµ (p′ .5 Form factors Currents play an important role in field theory. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P.3: The behavior of the independent bilinear combinations of fermi fields under P. γ µ . p′ |V µ (0)|p = u(p′ )γ µ u(p).38) (8. C. (q 2 ).g.37) (8.39) where Γµ (p′ . In many applications the expectation values of currents are needed. F. C. As an example consider the vector current for a point fermion. called form factors. Note that the lagrangian density L (x) → L (−x) under Θ. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. σ µν .. 8.36) (8. and T. p) can be built from any combination of Dirac matrices (1. The 16 combinations of Dirac matrices appearing above are linearly independent. (8. of which the expectation value between momentum eigenstates can be simply found. (8. 2M (8. 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).

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

0. Exercise 8. 4M 2 = F1 + F2 . 0.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. Exercise 8. (b) Show that such a term is also not allowed by time-reversal symmetry.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 .3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 0. (c) Show that if a term of this form would exist. 0. GE GM = F1 − Q2 F2 . −|q|/2). (b) Show that in the interaction with the electromagnetic field. σµν q ν F3 (q 2 ) 2M . p′ = (E. 0. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . |q|/2) with E 2 = M 2 + |q|2 /4 and express them in terms of the Sachs form factors depending on Q2 = −q 2 . (c) Show that hermiticity of the current requires that F1 en F2 are real. 0. |q|). p = (E.

t ′ QH (t′ )|q ′ . t0 ). t0 ) AH (t0 ) U (t. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t ′ |q . i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t′ ≡ q ′ |q ′ .1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian.2) (ii) Heisenberg picture. t0 ) = H U (t. ≡ 0. Consider the two (time-independent) Heisenberg states: |q. (9. in which the states are time-independent.Chapter 9 Path integrals and quantum mechanics 9. t0 ) ∂t (9. QH (t)|q.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. in which the operators are time-independent. AS (t) = AS and the states o are time dependent. t ≡ q|q.5) (9. (9.7) 73 . |ψS (t) = U (t. AH (t) = U −1 (t. (9. and the operators are time-dependent. t′ . (9. [AH . t0 )|ψS (t0 ) . |ψH (t) = |ψH .1 Time evolution as path integral U (t. t . i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0.3) (9.4) ∂ U (t. H].

qj ).. use the Campbell-Baker-Hausdorff formula. H(P.q . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . tj+1 |qj . 2 12 12 (9. B] + . 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 . QS |q ′ ≡ q ′ |q ′ . .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 ′ .12) where the transformation between coordinate and momentum space involves q|p = ei p. . [A. t′ |q. which is a hamiltonian function in which the operators are replaced by real-numbered variables.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . 2π (9. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 .. B] + [A. ′ (9.16) this. 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 . (9. qj )]) . . B].qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj .qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . .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 . Q) expressed in terms of the operators P and Q.11) (9. eA eB = eC with C = A+B+ (9. B]] + [[A. n |q (9. tj+1 |qj .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.14) . (9. Combining the two terms gives qj+1 . Then qj+1 .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . 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|. ˙ 2π 1 1 1 [A. The hamiltonian is an operator H = H(P.8) Choose t as the starting point with |q = |q. t′ |q. 2π (9. t = q ′ |e−i H(t −t) |q . The purpose of this is to calculate the evolution for an infinitesimal time interval. with the correction1 being of order (∆τ )2 . |q1 q1 |e−i H∆τ |q . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj .

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

23) Having defined the Gaussian integral.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. Using method 1 one writes 1 Dα exp − α2 2 = x dαx N exp − 1 2 ∆x α2 x x = x dαx 1 exp − ∆x α2 x N 2 1 N 2π ∆x ≡ 1. A useful property of functional integration is the translation invariance. (9. .28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .25) (see Exercises).21) again a multidimensional integral. In both cases N is an appropriately choosen normalization constant in such a way that the integral is finite. the following integrals can be derived for a real symmetric or complex hermitean kernel K.24) (9. N (9.22) = x The last equality is obtained by defining the right measure (normalization N ) in the integration. α(x) = n αn fn and K(x. Dα F [α] = Dα F [α + β]. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. 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. 2 (9. y) = ∗ m. Note that procedure (i) is an example of the more general procedure under (ii). 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. with Dα F [α] = n dαn N F (αn ) (9. Consider the gaussian functional as an example. (9.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .27) (9.

however. δα(x) (9. (9. In the expansion of the exponential (from second to third line). Examples of functional differentiation are (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.34) (9.35) For applications to fermion fields. F [θ. dx dy θ∗ (x) K(x. α(y) = δ(x − y). y) θ(y) .38) . products of the same pair appear. δα(x) δα(x) δ αKβ = dy K(x. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables.e.30) (9. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1.g.Path integrals and quantum mechanics 77 Functional differentiation is defined as δ .33) (9. θ2 = 0. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .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.31) δ δ α = dy α(y) = dy δ(x − y) = 1. e. θη = −ηθ.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 ) = 1.32) (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. which vanish. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). In principle the definitions of functionals is the same. i. (9. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . n θn fn (x) dx θ∗ (x) θ(x) . αn = δmn . the Gaussian-type functionals.29) (9. δα(x) or in discretized form δ 1 ∂ ∂ with . αy = δxy . (9.

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

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

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

det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . 2 (9. y) K −1 (y.78) (9. z) = δ(x − z).79) (9. where dy K(x. .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 .

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

As discussed in the previous section (exercise 9.6) the generating functional is (using ∂µ = − ∂µ ) given by   Z0 [J] =  Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). J(xn ).9) i. xn ) = x2 xn .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .10) J=0 In order to determine ∆F .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.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. (2π)4 (k 0 + iǫ)2 − E 2 (10. xn ) J(x1 ) . Furthermore.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) . . . (10. . 2 2 → (10. . with Z0 [0] = 1.11) Depending on the path choosen in the complex k 0 plane (see exercise 7. d4 xn G(n) (x1 . . . . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . .e.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. (10.8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10. . .

(10. 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). ∆(x) = − d4 k e−i k·x . the advanced Green’s function ∆A (x) = satisfying ∆A (x. (2π)4 k 2 − M 2 (10. (10. t) = 0 for t > 0. ∆F = ∆C . The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).e.18) Thus we see various solutions. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x).16) C where C is a closed contour in the complex k 0 -plane. (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 . µ 1 1 • Firstly.22) (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. φ+ (0)] = (2π)3 2E are also shown in exercise (7. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . t) = 0 for t < 0.4). the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . 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. 10.15) e−i k·x d4 k . 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 .19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) .14) They are solutions of the inhomogeneous equation. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). 4 k 2 − M 2 + iǫ (2π) (10. The first possibility is to consider the well-defined Euclidean formulation. i. (2π)4 (k 0 − iǫ)2 − E 2 (10. .17) d4 k e−i k·x .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.20) = exp − 2 where or (see fig.

x2 . or diagrammatically (4) (10. the same result as above. φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). = θ(x0 − y 0 ) 0|[φ+ (x). x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. x3 .1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. 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.24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function.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). We find (see also section 7.25) x1 x2 x3 = x4 + + . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . .Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. we have G0 (x1 . because this is the only quantity entering Z0 [J]. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F .

30) and in zeroth order G(0. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar.27) When interactions are present.Feynman diagrams for scattering amplitudes 89 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 . x4 ) = G0 (x1 .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. (10. Z[J] = = Dφ exp i exp i d4 z LI (10. x4 ) = To first order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. x2 . with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].g ) (x1 . (10.g 0 ) = G0 = 1. x2 ) = G0 (x1 .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 . L (φ) = L0 (φ) + LI (φ).g ) (x1 . (10. the generating functional can be written.e.29) This expression will be the one from which Feynman rules will be derived. x3 .35) z .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. i. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .34) Introducing a vertex point to which four propagators are connected. x2 ) = i ∆F (x1 − x2 ) = G(4. x3 .32) (10. (2) (0) (10. ≡ −i g d4 z (10. x2 .31) G(2.

δJ(xn ) .36) J J from which one obtains G(0. In the free case we have seen that the exponent of Z[J] contained all essential information. Connected Green’s functions We have now seen the Green’s functions as the quantities appearing in the expansion of the generating functional Z[J]. J(xn ). x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . (10. . This remains also true for the interacting case. . . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . . . did not contain parts that could be written as products of simpler Green’s functions. (10. d4 xn G(n) (x1 . .38) The result for the 4-points Green’s function is left as an exercise.40) in−1 n! d4 x1 . xn ) J(x1 ) . (10. . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . . . 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 . . . . Diagrammatically this exponent only contained the two-point function. . (10. . . J(xn ) c (10.41) in terms of socalled connected Green’s functions. . which also was the only Green’s function for which the diagram was connected. . .39) We have calculated the function up to first order in the coupling constant for φ4 -theroy. .37) G(2.Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J].g 1 ) = 1 8 z 1 8 . Z[J] = n in n! d4 x1 .g ) (x1 . (10.42) x1 G(n) (x1 . G(n) (x1 . . xn ) J(x1 ) . . . i. d4 xn G(n) (x1 . .e. .

(ii) Note that Z[0] = exp(iW [0]) appears as a multiplicative factor. 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). In the expansion of Z[J]/Z[0] one (by definition) has the socalled source (n) connected Green’s functions Gsc . If i W [J] = then exp(i W [J]) = 1 + i W [J] + = 1 + 1 2! +i d4 x1        i2 + d4 x1 d4 x2 2!         i2 4 4 d x1 d x2 +   2!     +                                        + (i W [J])2 (i W [J])3 + + .. 1 2! Z[0] = 1 + + + . + +i d4 x1        1 2! + 1 3! +i d4 x1 + 1 2! i2 2! d4 x1 d4 x2              + .. in the expansion of which all vacuum blobs are contained. 2! 3! +i d4 x1 + i2 2! d4 x1 d4 x2 .. (1) (1) (iii) Gsc = Gc .Feynman diagrams for scattering amplitudes 91 To see that Z[J] = exp(i W [J]) contains all non-connected diagrams can be performed inductively. (iv) If δZ[J] = i 0|φ(x)|0 = 0.... but is easily illustrated by writing down the first few terms. δJ(x) J=0 .

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

φ(y)] = Z [φin (x). e. this can actually not be used as it would imply [φ(x). a)SU −1 (Λ.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 ). √ Although the above.and out-fields is: φin (x) = S φout (x)S −1 . Proof: β. a) = S. The time evolution operator (see Eq.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. β. as it stands. The relation between S and Z[J] To establish this relation.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).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) . in|φin S → φin S = S φout (x). The convergence therefore must be weakened to t→−∞ √ (10. Considering plane waves one sees from √ √ 0|φ(x)|p = lim Z 0|φout (x)|p = Z e−i p·x t→∞ √ √ = lim Z 0|φin (x)|p = Z e−i p·x t→−∞ ↔ that identical normalization of (single-particle) plane waves in initial and final state implies the same wave function normalization Z for in and out fields.54) . S is invariant under e Poincar´ transformations: U (Λ. (10. 2 ∂µ ∂ µ + M0 φ(x) = J(x). The relation between S-matrix and in. 9.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . (2π)3 2E (10.48) The fact that φin and φout satisfy the homogeneous Klein-Gordon equation with the physical mass M implies that they create particles and antiparticles as discussed.51) (10.g. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. φin (y)] = iZ ∆(x − y). z) φ(z). which in a later stage (renormalization) will become more important. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance.52) where Z is a constant. implies the strong (operator) convergence φ(x) → Zφin (x). out|φout = β. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10.e. the source term Jφ in the lagrangian density is treated in the interaction picture. i. z) φ(z) d4 y d4 z ∆A (x − y) K(y. a causality condition that can be proven to imply the absence of interactions. (10. (10.

(10. δJ(z) (10. Thus applied to the field φ(x) = φ+ (x) + φ− (x). φ− f (10.56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits. we can. B] (for the case that [A.63) .e. y) δ δJ(y) : F [J]. provided [A. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.62) where F [J] is some (arbitrary) functional. B] is a c-number). which is equal to the expression e in which the creation operators are placed left of annihilation operators. S U [J]] = √ Z δ S U [J]. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. B]eB . : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A. t i Ut∞ [J] φ(x)U−∞ [J] (10. 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. eB ] = [A. one has [A. z) d4 y d4 z ∆(x − y)K(y. φ(y)] f (y) e e . B + C]e e . (10.Feynman diagrams for scattering amplitudes 94 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. φ(x).58) In order to find a solution to this equation note that. just as we did for φ(x). B] and [A. e = i. C] are c-numbers.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) .60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. z) δU [J] . δJ(z) Taking the difference between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. φ(x). e e ] = [A.59) (10. express it in terms of advanced and retarded Green’s functions.57) i [φin (x). [A. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. δJ(z) (10.61) e φ+ f where the normal ordered expression : e φ f : is used.

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

27. ’t Hooft and M. Veltman.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. or calculating full Green’s functions external lines are treated as propagators. 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 10.69) LI (φ) = − φ3 − φ4 . At these vertices each line can be assigned a momentum. vertices and external particle wave functions in that diagram. 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. to be precise iLI vertices in momentum space are introduced. Diagrammar) in the case of the interaction terms g f (10. Note that in calculating amputated Green’s functions external lines are neglected. For the symmetry factor consider the examples (given in G.27. = −i g . the factor 2 corresponds to the two-points Green’s function having two identical ends).66 and 10. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically different connected diagrams. but overall momentum conservation at a vertex is understood. 3! 4! The vertices are: = −i f Consider first the lowest order self-energy diagram. (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). = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). which cancels the factor 1/2 in the quadratic piece. For the interaction terms in the lagrangian. If problems arise go back to the defining expression for the generating function in 10.

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

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

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

s1 ) ¯ −igµν u(p′ . s)γν u(k ′ . s2 )|2 ¯ q2 (10. s′ ) ¯ u s.s2 .75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .79) .Feynman diagrams for scattering amplitudes 100 10.77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s′ )γµ u(k. s2 ). where u ¯ L(m) µν = = 1 2 u(k ′ .5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon.s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10. s3 )(−ie)γ µ u(k.s4 e4 (m) µν (M) L L . t2 (10.76) 1 . ¯ q2 (10. s3 )γ µ u(k. s1 ) u e2 u(p′ . q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. To lowest order (∝ α = e2 /4π) only one diagram contributes. If we are interested in the scattering process of an unpolarized initial state and we are not interested in the spins in the final state we need |M |2 summed over spins in the final state ( s3 . s4 )(−ie)γ ν u(p.s3 .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . s)¯(k. s4 )γµ u(p. The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ .

s4 = 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 . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. s3 )γ µ v(p′ . This is known as crossing symmetry. To lowest order (∝ α = e2 /4π) only one diagram contributes. s1 ) u(k.2.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s1 )|2 ¯ s 1 .s2 . s3 ) ¯ ¯ 1 . Similarly. . the masses. x4 ) for the interacting case in φ4 -theory to first order in the coupling constant g as obtained from the full expression for Z[J] in section 10. . s2 )γµ u(k. Exercises Exercise 10. s4 ) v (p′ . (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. (4) (4) . . s3 )γ µ v(p′ . s2 (10.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . s1 )γν v(k ′ .s4 = 1 2s × v (k ′ . s4 ) u e2 v (k ′ . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. 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.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. The diagram.s3 . (c) Do the same for the connected Green function Gc .s2 3 . s4 )γ ν u(p. s2 )γµ u(k. s2 ) ¯ ¯ 1 2s u(p.

. . Pictorially this implies. It is of the form −i M = A1 − A2 . Note that the function G(n) (p1 . . which means overall momentum conservation in Green functions in momentum space. xn + a) = G(n) (x1 . Express the contributions in invariants s. pn ) only has n − 1 independent variables in it.3 Show that the combinatorial factors found using the rules given in section 10. . .Feynman diagrams for scattering amplitudes 102 Exercise 10. . Exercise 10. . . Exercise 10. . . or if we in general would consider all momenta as incoming. . 1 2 (b) Calculate the quadratic pieces and interference terms. pn ). but the incoming and outgoing momenta are identical.4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. . for instance.2 (a) Show that the translation properties of the fields and the vacuum imply G(n) (x1 + a. + pn ) G(n) (p1 . . in the amplitude (Tij = A∗ Aj ). . . hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . . their sum is zero. . e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . . . . xn ) ≡ (2π)4 δ 4 (p1 + . . . . . |M |2 = T11 + T22 − T12 − T21 . Show that i the amplitude is symmetric under the interchange of t ↔ u. . xn ). t and u. . .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10.37 and 10. xn ) ∝ (2π)4 δ 4 (p1 + .38. 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). + pn ). .

. . (11. . In that case cm pa = (Ea . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . i n d3 pi . s = (p1 + .3) which is useful because the total 4-momentum of the final state usually is fixed by overall momentum conservation. . Its square root. Here s is the invariant mass of the n-particle system. For two particles. . In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . p1 .Chapter 11 Scattering theory 11. pa ) and pb = (Eb . + pn )2 . . . 103 . . pb ) satisfying p2 = m2 and p2 = m2 . (11. the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ). (11. s is for obvious reasons known as the center of mass (CM) energy. To be specific let us consider two frequently used frames. It is a useful quantity. and the total momentum four-vector P = pa + pb . −q). pn ).7) pb = cm (Eb . The phase space is determined by the weight factors assigned to each state in the summation or integration over states. (2π)3 2Ei i=1 (11. (11. . It is determined by the energy-momentum four vector p = (E. . This is generalized to the multi-particle phase space dR(p1 . . pb we start with the four vectors pa = (Ea .5) √ is referred to as the invariant mass squared. i.2) (11.e. . a a b b the quantity s = P 2 = (pa + pb )2 . A physical state has positive energy. . q).6) (11.1) = (2π)3 2E (2π)4 (proven in Chapter 2). . The first is the CM system.4) For two particle states |pa .1 kinematics in scattering processes Phase space The 1-particle state is denoted |p . for instance for determining the threshold energy for the production of a final state 1 + 2 + . pn ) = and the reduced phase space element by dR(s. . . p) which satisfies p2 = E 2 −p2 = m2 . + n.

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. which in the specific case a b also can be expressed as λ(s. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. Also in this case one can express the energy and momentum in the invariants. 2 s s − m2 + m2 a b √ . 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 √ = . 2mb (11. m2 ). 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 √ = . ptrf ). m2 . = 2 s λ(s.13) (11. a (11. m2 . mb and s).10) The function λ(s. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 0).11) (11. m2 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 . m2 ) a b .14) One can.9) (11.15) Explicit calculation of the reduced two-body phase space element gives dR(s. p1 . 2mb λ(s. p2 ) = CM = = 1 (2π)2 1 (2π)2 1 (2π)2 d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2 which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s. 4s (11. p1 . 4π s 4π 8π s 4π (11.8) (11.12) pb = (mb . m2 ) a b . In that case trf pa = (Ea . for instance. (11.16) . p2 ) = = where λ12 denotes λ(s. m2 ) is a function symmetric in its three arguments. m2 ) = 4(pa · pb )2 − 4p2 p2 .

23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . pz = mT sinh y. In inclusive measurements no particles are detected in the final state. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.. pT = (px . 2s (11. E . in which one particle is detected.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. Of course there are not only 2-particle channels.19) (11. py ). usually many particles are produced. = (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees.27) with m2 = m2 − p2 = m2 + p2 + p2 .17) (11.22) (11. 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 .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. The various possibilities are referred to as different reaction channels. Consider the process a + b → c + d. (11.21) (11. pc . Writing E = mT cosh y. 11.18) (11. the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . An often used set of invariants are the Mandelstam variables. tmax min λcd /4s. Consider the 1-particle inclusive case. while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). 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 . Instead of the scattering angle. however. where the first is referred to as elastic channel and the set of all other channels as the inelastic channels. H1 + H2 → h + X. At high energies. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11.. A specific reaction channel starts contributing at the threshold value (Eq. The variable s is always larger than a c b d the minimal value (ma + mb )2 . s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. pd ) = 4π s 4π 8π s 4π dt dt √ √ . .24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. usually is a 2-particle state.4). there is usually a preferred direction.25) dR(s. Examples appear in → π− + p (11. The initial state. in exclusive measurements all particles are detected.20) → π0 + n → π+ + π− + n → .

For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. b ut = (pa − pd )2 = u. t = (p1 − p3 )2 = (p2 − p4 )2 . ¯ tt = (pa − p¯)2 = s. uu = (pa − p¯)2 = s. (11. uu = s) = Mab→cd (s. (11. t. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. ¯ b (11. This is known as crossing symmetry. t. η = − ln(tan(θ/2)) = tanh−1 (cos θ). For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. ut = u) = Mab→cd (s. u = (p1 − p4 )2 = (p2 − p3 )2 . ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. 11. b Analiticity of the field theoretically calculated result implies Ma¯→¯ (st = t. With the momenta defined as in the figures above one has for all these processes the same amplitude M (s.2 Crossing symmetry In the previous chapter.31) . 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. u) with s = (p1 + p2 )2 = (p3 + p4 )2 .Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. t-channel and u-channel processes respectively. ¯ tu = (pa − pc )2 = t. tu = s.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1.28) which (only involving angles) is easier to determine. which means that rapidity distributions dN/dy maintain their shape. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd).30) (11. u). t. u). 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 = = = . tt = s. c bd Mad→c¯(su = u.

Scattering theory 107 One can also phrase it in the following way: one has a single analytic function Mab→cd (s.32) σ= trf T · V ρa ρb va co . This is a consequence of the constraint 2 2 2 2 s + t + u = Ma + Mb + Mc + Md . 2 2 2 2 u(t − s) + (Ma − Md )(Mc − Mb ) . . while the flux of the beam particles a is trf flux(a) = ρa · va . _ 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 · flux(a).e. t and u. (consider for convenience the rest frame for the target. To see this one can make a two-dimensional plot for the variables s. . i. The proportionality factor has the dimension of area and is called the cross section. This defines the boundaries of the physical regions. u) that represents physical amplitudes in the physical regions for three scattering processes. 1 N .) is defined as the proportionality factor in Nc = σ(a + b → c + . . Nb indicates the number of (target) particles b. (11.3 Cross sections and lifetimes Scattering process For a scattering process a + b → c + . say b) the cross section σ(a + b → c + . T where V and T indicate the volume and the time in which the experiment is performed. Nc /T indicates the number of particles c detected in the scattering process. t. . . . which for a density ρb is given by Nb = ρb · V . shown below for the case of equal masses. 2 2 2 2 t(s − u) + (Ma − Mc )(Mb − Md ) . To find the physical regions one looks for the boundaries of cos θs = cos θt = cos θu = 1 λs λs ab cd 1 λt λt ac bd 1 λu λu ad cb 2 2 2 2 s(t − u) + (Ma − Mb )(Mc − Md ) .

e. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . . The decay time for moving particles τ is related to the decay time in the rest frame of that particle (the proper decay time τ0 ) by τ = γ τ0 . . (11. σ= √ 2 λab T · V (11. N 1 .35) This quantity is not covariant.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .36) (11. . (11. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. For the (proper) decay width one thus has 1 N Γ0 = . 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 (11.37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears. . T · V ρa (11.T V.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). N and T · V are covariant. as expected. pn ). it is.38) (in which we can calculate −iMf i using Feynman diagrams).33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ.34) i. dt (11.Scattering theory 108 Although this at first sight does not look covariant. (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. . 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 .

θ) = −8π s ℓ (in analogy with the expansion for f (E. (11. (11. θ)|2 .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. (11. respectively. 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 the √ amplitudes squared have been calculated in the previous chapter. 11. q f ). The result is N = |Mf i |2 dR(s. (11. p1 . 2 λab and for instance for two particles dσ = q ′ M (s. q n ). i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . θ) in quantum mechanics. t) λab 4π dt.47) M (s. q n ) q n qn √ Mni (q i . pn ). .46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. . . . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM.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 . . .42) 2−body decay The differential cross section (final state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. 8π s 2π 8π s . see 11. . (11. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . . ˆ ˆ Inserted in the unitarity condition for M . p1 .Scattering theory 109 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). θcm ) √ q 8π s 2 2 (11.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . pn ). note the sign and cos θ = qi · qf ). implies for the scattering matrix M . i. The sign difference comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. . pn ) |M |2 q 32π 2 m2 dΩ |M |2 . p1 . for which one has dσ/dΩ = |f (E.e. θ).41) (11. .43) dΩcm = π M (s.44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. θ) or M (q i . 16π 2 s (11. .

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

in which case. such that |U (t.58) . η = 0. 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. s − M 2 + iM Γ (11. For an unstable particle one expects that U (t. i. This is what we will do to discuss decay widths. 0)|2 = e−Γt . . ∝ U (t. The quantity Γ is precisely the width for unstable particles. 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. To check unitarity we need the amplitude at order e4 .Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . i. We note that unitarity is generally broken in a finite order calculation. . σel = σinel . 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 . This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. (q Mℓ )ij (s) = −M Γi Γj .55) (note that we have disregarded spin).e.e. specifically one has for t > 0 that the Fourier transform is dk e 0 −ik0 t ∆(k) ∝ e−ik t dk (k 0 − Ek + iǫ)(k 0 + Ek − iǫ) 0 0 (t>0) ∝ e−iEk t . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. the time-evolution operator.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. which is given by a sum of the partial widths into the different channels. . 0). (11. . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. 0) ∝ e−i(E−iΓ/2)t .57) This shows that Γ is the width of the resonance. it is straightforward to write down the partial wave amplitude. For instance the tree-level calculation of Feynman diagrams at order e2 was real. 11. The product of amplitudes is of order e4 . The prescription for the pole structure. For the amplitude in a scattering process going through a resonance.56) (again disregarding spin). however.

dt s(s − 4m2 ) with f (t. At resonance the cross section σT (π + p) is about 210 mb. This is a narrow resonance discovered in 1974. Its mass is M = 1232 MeV.2 GeV. the partial width into e+ e− is Γee = 5. i.49 GeV. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . which is attributed to neutrinos.e.26 keV. From this one sees that a resonance has the same shape in all channels but different strength. u) + g(t. All these decays can be seen and leave an ’invisible’ width of 498 MeV. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . Exercises Exercise 11. The half-width of the resonance is M Γ. Limiting ourselves to a resonance in one channel. Three neutrinos explain the resonance shape. a fast change in the phase shift for a narrow resonance. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. 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).4) can be written as 4πα2 dσ = {f (t.Scattering theory 112 (Prove this using the unitarity condition for partial waves). The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. the full width is Γ = 88 keV. its width Γ = 120 MeV.1 Show that the cross section for electron-electron scattering (exercise 10. Γ = 2.88 MeV. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering.59) reaching the unitarity limit for s = M 2 . Knowing that each neutrino contributes about 160 MeV (see next chapter). • The J/ψ resonance in e+ e− scattering. 2 q (s − M 2 )2 + M 2 Γ2 (11. t)} . M2 − s (11. t) + g(u. u) = g(t. • The Z 0 resonance in e+ e− scattering with M = 91. where furthermore σinel = 0. Its mass is M = 3096. This characteristic shape of a resonance is called the Breit-Wigner shape. u) + f (u. The cross section at resonance is about 30 nb. Note that because of the presence of a background the phase shift at resonance may actually be shifted. one can reconstruct the resonance shape for different numbers of neutrino species.

s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Exercise 11.8. ρcm ρcm a b 1 cm .Scattering theory 113 Exercise 11. Calculate now fπ . Exercise 11. 4.4 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∆ . a quantity which we will ¯ ¯ encounter in the next chapter (section 12.2 Calculate the flux factor. we have here already used unitarity to find a finite imaginary part that is proportional to the total decay width. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.999 877 Γ and Γ(π − → e− νe ) = 0.000 123 Γ. Show that unitarity can be used one step further to fix also the real part of the residue of the pole in the amplitude. As discussed for unstable particles. Thus.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. cm |va − vb | in the center of mass system for two colliding particles a and b. writing N (s) M1 (s) = 2 + iM Γ . Compare the results with the experimental π − lifetime τ = 2.2).

Chapter 12

The standard model
12.1 Non-abelian gauge theories

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

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

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

(12.3)

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

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

The standard model

115

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

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

The standard model
xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

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

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

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

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

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

    λ7 =   

i

    −i   

  1  1  λ8 = √   3

−2

However. ds which is solved by dψ(s) ds = dxµ .14) (12. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). (12. = . i. 12. i. ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). but expresses them with respect to different bases.16. should be independent of such transformations. which connects two identical vectors. In principle such a phase in a given point is not observable. A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.19) This gives rise to a (path dependent) phase in each point. if two different paths to the same point give different phases the effects can be observed. -dx -dy µ µ dyµ µ dx For a vector. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. The relation in Eq.e. If there is a ’true’ difference in the vector ψ(x) and ψ(x + dx) it is denoted with the covariant derivative connecting the vectors expressed with respect to the same basis.e. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).The standard model the basis for x G and x+dx 117 G. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). (12.15) ψ(x).17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). (12. (12. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.18) or considering a path xµ (s) from a fixed origin (0) to point x. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx). We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). 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. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj .    φ1    φ .2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. 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). a theory for a scalar degree of freedom (a scalar field) having three (real) components. external magnetic fields. .2. Nature will choose one.2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. 12. If it is zero one has dxµ Aµ (x) = 0. we can disregard those ’perturbations’ and look at the ideal situation. and equivalently Aµ can be considered as a pure gauge effect.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 .The standard model 118 where Gµν = (i/g)[Dµ . e. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm effect). In this case there are many possible groundstates (determined by a fixed direction in space). usually being (slightly) prejudiced by impurities. As there can be one and only one groundstate. the groundstate itself appears degenerate.e. in reality a not perfectly symmetric situation. Dν ]. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. i. 2 2 4 −V (φ) The potential V (φ) is shown in fig. i>j which is rotationally invariant. Nevertheless. this means that there is more than one possibility for the groundstate.  φ= 2      φ3 (12. Only if this quantity is nonzero a physically observable effect of Aµ exists. If the temperature is below a certain critical temperature (T < Tc ) the kinetic energy is no longer dominant and the above hamiltonian prefers a lowest energy configuration in which all spins are parallel. In geometric language the effect on parallel transport of a vector depends on the ’curvature’ Gµν . This characterizes spontaneous symmetry breaking.g.

Writing the field φ as a sum of a classical and a quantum field. (12.25) 2 2 2 Therefore there are 2 massless scalar particles. The η massless particles are called Goldstone bosons. i. while the lagrangian including the nonzero groundstate expectation value chosen by nature. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . A quantum field theory has only one nondegenerate groundstate |0 . In the case of degeneracy.e. The field along the third axis plays a special role because of the choice of the vacuum expectation value. In this situation one speaks of spontaneous symmetry breaking. has less symmetry.The standard model 119 a constant field (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2. The classical groundstate appears degenerate. The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized. while the quadratic terms must correspond with non-negative masses. In this case the spectrum is described . In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . therefore a choice must be made. . . corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . . 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 (12.23)     F +η the latter only forming a minimum for m2 < 0. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . It is appropriate to redefine the field as    ϕ1      φ =  ϕ2  . It is only invariant under rotations around the 3-axis. for the vacuum Qa |0 = 0. (12. The situation now is the following. Any constant field with ’length’ F is a possible groundstate. say    0    0 .) terms constitute interaction terms. ϕ2 and η. etc.21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero). It is sufficient to do this to second order in the fields as the higher (cubic. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|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.   ϕc = 0|φ |0 =  (12. They are known as the Weyl mode or the Goldstone mode: Weyl mode. (12. φ = ϕc + φquantum where for the (operator-valued) coefficients in the quantum field one wants 0|c† = c|0 = 0.

Taking the derivative.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). V µ = ψγ µ T ψ.    .e.e. (12. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). The generators Qa (in that case the rotation operators Lz .The standard model 120 by degenerate representations of the symmetry group. and one must have mπa = 0. Goldstone mode.. . This means that they are operators that create states from the vacuum. T ] ψ. Including a sum over the different flavors (up. there must be a nonzero expectation value 0|Jµ (x)|π (k) . A bit more formal. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . i. i.26) for all the states labeled by a corresponding to ’broken’ generators. ∂ ∂ (12. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. H] = 0.  . M =      .    . a massless Goldstone boson for each ’broken’ generator.32) (12.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. denoted |π a (k) . [Qa . = ψ(i/ − M )ψ. called flavors. irrespective of the fact if they annihilate the vacuum. T = τ /2. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. Lx and Ly or instead of the latter two L+ and L− ) are used to label the multiplet members or transform them into one another. if the generators Qa generate a symmetry. a ∂ µ Jµ = 0. Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab (fπ = 0) (12. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. Note that for the fields π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . Using the Dirac equation. (12.) one can write L = f ψ f (i/ − Mf )ψf . This lagrangian density then is invariant under unitary (vector) transformations in the flavor space. (12. i. it is easy to see that one gets ∂µ V µ = i ψ [M. a and a′ are degenerate states.30) which in the case that we include only two flavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. strange.31) . π (12. ψ −→ ei α·T ψ.28) where ψ is extended to a vector in flavor space and M is a diagonal matrix in flavor space. down. etc. Neglecting at this point the local color symmetry. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m).      ψu   mu           md     ψ =  ψd  .e.

From group theory (Schur’s theorem) one knows that the latter can only be true.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . while the groundstate is only invariant under isospin rotations (R = L).27. or. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. parity minus).e. if in flavor space M is proportional to the unit matrix. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. both are very small. This situation. etc. parity would not play a role in the world. 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. Aµ = ψγ µ T γ5 ψ. is what happens in the real world. i. M = m · 1. where the quark masses are not completely zero. with almost degenerate masses (Mp = 938. is by nature not realized in the Weyl mode. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . ψ −→ ei α·T γ5 ψ. The chiral fields ψR and ψL are transformed into each other under parity. From the number of broken generators it is clear that one expects three massless Goldstone bosons.e.e. Therefore the world of up and down quarks describing pions. generated by TR/L = 1 (1 ± γ5 )T . ψR/L = 1 (1 ± γ5 )ψ. but the fact that the symmetry is not perfect and the right hand side of Eq.3 MeV/c2 and Mn = 939.6 MeV/c2 ).The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. . T } γ5 ψ. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. however. it is easy to see that one gets ∂µ Aµ = i ψ {M. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. stated equivalently. (leaving out the flavor structure) the same as ψγ5 ψ. much smaller than any of the other mesons or baryons. a doublet (isospin 1/2) of nucleons. proton and neutron. which is equivalent 2 to the V-A symmetry. We know that mesons and baryons (such as the nucleons) have a well-defined parity that is conserved.33) which in the case of two flavors form SU (2)A transformations generated by the Pauli matrices.g. i. ∂ ∂ (12. T ] = 0. 12. (12. (12. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . I. There exists another set of symmetry transformations.35 is nonzero. for which the field (according to the discussion above) has the same behavior under parity.34) Using the Dirac equation. In the real world. but also a triplet (isospin 1) of pions. behaves as a pseudoscalar particle (spin zero.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. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. socalled axial vector transformations. including in Dirac space a γ5 matrix. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. T γ5 = τ γ5 /2. e. which is approximately true for the up and down quarks.e. This symmetry. fπ = 0 remains. chiral symmetry is not perfect. Note that these transformations also work on the spinor indices. 12. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . (12. the lagrangian density is invariant under left (L) and right (R) rotations independently. How can we see this. etc.e. This combined symmetry is what one calls chiral symmetry. i. as the quantity ∂µ Aµ (x).

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). After symmetry breaking the same number (as expected) comes out. . 4 2 where Since the explicit (adjoint. made into a gauge theory. while the number of massive gauge bosons coincides with the number of ’broken’ generators.The standard model 122 12. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. (12. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar field (Higgs field) with three components. . is made. Initially there are 3 massless gauge fields (each. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). One may wonder about the degrees of freedom.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     . . We have the possibility to perform local gauge transformations.    0      ϕc = 0|φ |0 =  0  . Therefore we can always rotate the field φ into the 3-direction in order to simplify the calculation.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. thus 9 independent degrees of freedom.    0 φ= (12. The difference comes when we reparametrize the field φ.42) 2 from which one reads off that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).e.   The symmetry is broken in the same way as before and the same choice for the vacuum. (12. one sees that the fields Wµ and Gµν also can be represented as three-component fields. (12. i. having two independent spin components) and three scalar fields (one degree of freedom each). like a photon. as in this case there are no massless Goldstone bosons. again 9 degrees of freedom. 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. but one has 1 massless gauge field (2).38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . 2 massive vector fields or spin 1 bosons (2 × 3) and one scalar field (1).39) (12.37) (12.40) a Dµ φ = ∂µ φ − ig Wµ La φ.     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 + . .     0  0         0 = .

The procedure. (12. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. By introducing gauge fields and breaking the symmetry a more complex lagrangian is obtained. U(1)Y U(1)Y (12. 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 = 0.48) (12.and righthanded particles. ψR/L = 2 (1 ± γ5 )ψ decouple. the leptons (elektrons.47) and YW is considered as an operator that generates a U (1)Y symmetry.The standard model 123 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.) and the gauge bosons responsible for the strong. i. the quarks (up. its corresponding neutrino (νe . This is written as 3 Q = TW + YW .51) .46) −→ 3 One notes that the charges of the leptons can be obtained as Q = TW − 1/2 for lefthanded particles 3 and Q = TW − 1 for righthanded particles. neutrinos. that gives a good description of the physical world. etc. however.e. while the lefthanded particles form a doublet.).43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. ∂ ∂ ∂ (12. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. = ∂µ L − (12. down. τ ). left. muons.44) which has an SU (2)W symmetry under transformations eiα·TW . 2 (12. etc. νµ and ντ ) and their antiparticles. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. R −→ e−i β R. muon µ− or tau τ − ).45) (12. electromagnetic and weak forces.50) (12. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . R.49) Next the SU (2)W ⊗ U (1)Y symmetry is made into a local symmetry introducing gauge fields Wµ and Bµ in the covariant derivative Dµ = ∂µ − i g Wµ · TW − i g ′ Bµ YW /2. (12. 2 2 = ∂µ R + i g ′ Bµ R. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. explicitly L −→ e−i β/2 L. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. 1 As they are massless.

57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .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. leading to the parametrization   1  0    φ(x) = √  (12. . −V (φ) L and (h2) = −Ge (LφR + Rφ L). + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . Putting this in leads to L (f ) =L (f 1) +L (f 2) .   h)  (12. 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 = 0.52) L L (f 1) (f 2) = = i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ + i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 . L where L (h1) (h) =L (h1) +L (h2) . For the classical field the choice θ4 = v is made. . Using local gauge invariance θi for i = 1.59) . a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )    0 = 1   (12. 4 4 In order to break the symmetry to the symmetry of the physical world. . this leads to the lagrangian L (h1)     . the U (1)Q symmetry (generated by the charge operator).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. + 8 (12. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). . .53) φ=   √ (θ4 − iθ3 )  φ 2 with TW = 1/2 and YW = 1 is introduced.56)  v + h(x) 2 Dµ φ = (∂µ − and     Dµ φ =     = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms.58) (12. (12. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. † i i g Wµ · τ − g ′ Bµ )φ. (12. which assures with the choice of YW for the Higgs field that Q generates the remaining U (1) symmetry. (12.

(12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. we can read off e = g sin θW = g ′ cos θW . The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ .66) From the coupling to the photon.61) (12.63) (12. 4 2 MW g2 v2 = .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 . 2 Zµ (12. 2 (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 . . (12.64) (12.69) . 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 + . 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 . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12. g ′ /g = tan θW .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .60) (12. 2θ 2θ 4 cos W cos W 0.65) The weak mixing angle is related to the ratio of coupling constants.67) The coupling of electrons or muons to their respective neutrinos.

such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. one has the relation g2 e2 1 GF √ = . (12. giving rise to a running coupling constant after renormalization of the field theory.e.166 4 × 10 0.g.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. me /v is extremely small.80) First. muon or tau.e. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). i. −5 (12. . For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.72) (12. 91. 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. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). The coupling to the Higgs particle is weak as the value for v calculated e.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 . 1.73) (12. from the MW mass is about 250 GeV. Finally we want to say something about the weak properties of the quarks. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. (12.76) with CV CA 3 = TW − 2 sin2 θW Q.74) (12. i. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). (12. Comparing this with the total width into (invisible!) channels. 2 2 2 (12. 80.5 MeV (exp.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. g ′ and v determine a number of experimentally measurable quantities. v (12.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 3 = TW .40 GeV.77) (12. g2 MZ 2 2 (CV + CA ).05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.71) GeV −2 = = = = . (12. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. With the choosen vacuum expectation value for the Higgs field.19 GeV. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.231 2.

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

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

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

hence the mixing matrix becomes   iα /2 0 0   e 1     (12. For these light Majorana neutrinos one has. a unitary matrix relating them to the weak eigenstates. one must for the neutrinos as well as charged leptons. Thus (disregarding family structure) one has two Majorana neutrinos. For the neutrino sector. 12. 1 ∗ c c (12. Thus 2 Υ ′ = nc + nL . coupled by a Dirac mass term.95) 2 In order to have more than a completely decoupled sector.98) (12. it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ.and righthanded species then just form a Dirac fermion. are equivalent to two decoupled Majorana neutrinos (see below). however. (12. one being massive.c.ν =− containing three (CP-violating) phases (α1 . For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates.The standard model 130 The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. α2 and δ).       0 0 1 L mass. We use here the primes starting with the weak eigenstates. 2 R ψ = nR + nL . is to add in Eq. The mass matrix can be written as   ML |MD | eiφ     M = (12.96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  . couple the right.ν = − MR NR NR + MR NR NR . −    MD MR nR 2 (12.99) . 2 although this option is not attractive as it violates the electroweak symmetry. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. 1 L Υ ′ = nR + nc . The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors.94) M L NL NL + M L NL NL . the left. For the leptons.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section.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. Absorption of phases in the states is not possible for Majorana neutrinos. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. as above. with for the righthanded sector a mass term MR . 1 c ∗ c L mass. (12. This is called the see-saw mechanism (outlined below).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. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). the massless and massive Majorana neutrinos. Actually.

Exercises Exercise 12. 2 with real masses Mi . 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 . 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′ .  (12.105) for i = 1.102) = 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)) .e. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † .      Υ2R nR (12. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates.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 .1 Consider the case of the Weyl mode for symmetries. Prove that if the generators Qa generate a symmetry.The standard model 131 taking ML and MR real and non-negative. (12.101) Thus one obtains from  2 ML + |MD |2  † MM =   |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2   . T ] ψ. i. 2 U (M M † ) U † = M0 . i ψ {M.  L      Υ2L nR Υ1 and Υ2 . V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. . a and a′ are degenerate states.e. [Qa . This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families). Exercise 12. Exercise 12. related via    c   Υ1R     = U  nL  .103) for each of which one finds the lagrangians L = and we are left with two decoupled Majorana fields      Υ1L    = U ∗  nc  .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12.2 Derive for the vector and axial vector currents. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 finds M1 ≈ MD /MR and M2 ≈ MR . T } γ5 ψ. (12. H] = 0. i.

ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). mb ≈ 5 GeV and MW ≈ 80 GeV. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) .6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. .4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. the weak mixing angle is given by ¯ sin2 θW = 0. and the width to a pair of neutrino’s. Γ(Z 0 → νe νe ). ZZh. hhh. Mf M2 and ghf f = . Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . Γ(Z 0 → e+ e− ). e+ e− h. Exercise 12. Exercise 12.) are proportional to the mass squared for bosons or mass for fermions of the particles. ghbb = b v v where v is the vacuum expectation value of the Higgs field. Note that you can find the answer without explicit construction of all interaction terms in the lagrangian. Exercise 12. etc.7 In this exercise two limits are investigated for the two-Majorana case. (b) Calculate the width to electron-positron pair. at least up to a high (which?) order in λ.231. Write down the matrix element squared (averaged over initial spins and summed over final spins) for the decay of the Z 0 . Exercise 12. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . 3 TW . to be precise one has for bosons (b) or fermions (f).5 Show that the couplings to the Higgs (for W + W − h.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . 2 2 The masses are mt ≈ 175 GeV. The mass of the Z 0 is MZ = 91 GeV.

(b) A more interesting situation is 0 = ML < |MD | ≪ MR .The standard model 133 (a) Calculate for the special choice ML = MR = 0 and MD real. Calculate the eigenvalues ML = 0 and MR = MX . Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 . . Given that neutrino masses are of the order of 0. what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV).05 eV. which leads to the socalled see-saw mechanism. 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.

Sign up to vote on this title
UsefulNot useful