This action might not be possible to undo. Are you sure you want to continue?

# 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 6. 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. Corresponding chapters in books of Ryder. Quantum Field Theory. Quantum Field Theory.3 2.2 36 22 .1 1.2 3.2 6.5 5.1 6.3 2. Vol.8 3.3 3. 1996.3.1 3.1 5.3 3. 2. 4.2 3.1 2 2 3.2 1.1 2. M.7 2. 1994. Ryder. Cambridge University Press. 5. These notes Section 1.3 2.3 2. The quantum theory of ﬁelds.V.1 3. M. Peskin & Schroeder and Srednicki.-B. Addison-Wesly. Schroeder. M.2 2. An introduction to Quantum Field Theory. Cambridge University Press. McGraw-Hill.2 2. 1. Cambridge University Press.2 4.1.3 References As most directly related books to these notes. 3.2 3.1 4. 1995.4 3. Cambridge University Press.H. 1980.2 3. Quantum Field Theory. The notes contain all essential information. Srednicki. 6. Weinberg. 1985. Cambridge University Press.4 2. Zuber. but are rather compact.3 4.3 3.5 4. 1995.3 6. 3. S.2 3.E. 2007.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. Other text books of Quantum Field Theory that are useful are given in refs [3-6].3 3. Diagrammatica.4 4.4 3. Peskin and D.1 2.2 2.2.3 3.4 2. Vol. I refer to the book of Srednicki [1] and Ryder [2].4 3. I: Foundations.3 2. L.5 2. 3. Itzykson and J. Veltman. II: Modern Applications.5 2.4 6. C.

2 7. 2. 6.3 2.6 4.4 These notes Section 7.6 6.4 4.3 10.1 12. 9.2 9.6 3. 8.3 12.3.2 11. 6.1 9. 4.2 4.5 9.3 20.3 9.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.6 9. 4.1. 4.2 8.7 5.8 6.1. 2.1 5.5 4.3.4 7.5 7.3 8.5 7.4 12. 6.2 8.5 9.6 3. 6.4.6 10.7.4.3 3.7 4.3 6.5 2.5 .1 20.1 8.5.1 20.6 8.2 5. 6.2 10.3.8 5 4. 9.1 20.2 12.4 8.2.2.5 6 7 6 6. 4.4 2.4 12. 20.5 6.1.3 8.10 8.1 10.4 3.5 11.3 11.1 7.2.1 9.5 4.6 3.4.4 10.2.3 7.3 7.3 6 8 9 Srednicki 3 3 3 37 5.1 11.4 9.

• Slight rewriting of discussion of Majorana mass term in section 6.2 .5 Corrections 2011/2012) • Extension of Exercise 3.3 • First two sentences of Chapter 5 have been modiﬁed.

the position can in principle be determined at any time with any accuracy. In quantum mechanics a special role is played by Planck’s constant h. being eigenvalues of the position operators.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. In special relativity a special role is played by the velocity of light c. c = 299 792 458 m s−1 . ˜ ψ(p) = i d3 r exp − p · r ψ(r).Chapter 1 Introduction 1.4) where δij is the Kronecker δ function. One can talk about states |r and the wave function ψ(r) = r||ψ . ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. (1. Corresponding position and momentum operators do not commute.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. It is given by pop ψ(r) = −i ∇ψ(r). the action of the momentum operator in the coordinate representation is not as simple as the position operator. 1 (1.054 571 68 (18) × 10−34 J s = 6. (1. S ≫ .2) In the limit that v ≪ c one reaches the non-relativistic domain. In quantum mechanics. quantum eﬀects do not play a role anymore and one is in the classical domain. (1. They satisfy the well-known (canonical) operator commutation relations [ri . In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r).5) One can also choose a representation in which the momenta of the particles are the dynamical variables. (1. usually given divided by 2π.6) . (1. 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. Indeed.582 119 15 (56) × 10−22 MeV s.1) In the limit that the action S is much larger than . ≡ h/2π = 1. pj ] = i δij .

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.

. xi are the three space components. where the coordinate t = x0 is referred to as the time component.2. Unlike in Euclidean space. We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. 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 . x1 .3). When 3-dimensional space is considered as part of Minkowski space. The (invariant) lengths often have special names. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t.24) The quantity gµν ≡ nµ · nν is the metric tensor. one must distinguish tensors with upper or lower indices and one can have mixed tensors. ˆ ˆ (1. . So we could have used all upper indices in the above equations.21) (1. are . linear transformations that do not change the length of a four-vector are called the Lorentz transformations. denoted as V ′µ = Λµν V ν . Λµn n T ν1 .. x1 . In Minkowski space.. we will use upper indices for the three-vectors. The length (squared) of a vector is obtained from the scalar ˆ product. Vectors are denoted x = xµ nµ . No diﬀerence is made between upper or lower. independent of a coordinate system. x) = (x0 .Introduction 5 that can be used in the cross product of two vectors z = x × y.26) (1. (1. x2 . Because of the diﬀerent signs occurring in gµν . xµ = gµν xν . (1. Many other four vectors and tensors transforming like T ′ µ1 . 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 points lie on the lightcone and they can be connected by a light signal. x3 ). and are given by (x0 . x3 ) = (t. The real. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations. The lower indices are constructed in the following way. One has x2 = xµ xµ = t2 − x2 . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). it is convenient to distinguish lower indices from upper indices..1. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. however. • ds2 < 0 (spacelike intervals). in which case zi = ǫijk xj yk .3) with basis nµ (µ = ˆ 0. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . One can distinguish: • ds2 > 0 (timelike intervals). In special relativity we start with a four-dimensional real vector space E(1. δim δjm δkm δin δjn δkn . The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .22) (1.25) Within Minkowski space the real. These transformations do change the components of a vector. The Poincar´ transformations include e e Lorentz transformations and translations. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . −x). the invariant length or distance (squared) is not positive deﬁnite.νn can be conν structed.23) δjm δkn − δjn δkm . Relations relating tensor expressions. • ds2 = 0 (lightlike or null intervals). x2 . 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 .µn = Λµ1ν1 . .

27) It is easy to convince oneself of the nature of the indices in the above equation. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . = ∂ . The length squared of ∂ is the d’Alembertian operator. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . . (1. a∞ = 4πǫ0 2 /me e2 .34) (1. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity.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. g0 = g1 = g2 = g3 = 1. constructed via the metric tensor itself. where m is called the mass of the system.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). Each of these permutations leads to a minus sign. because one has (1. Note that in this equation one has on left. 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. p) = (E.Introduction 6 called covariant. 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 . but more important has the same eﬀect on lefthandside and righthandside. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .32) . p). ∂t (1. . ∂1 . It is an invariant tensor.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . ∂3 ) = ∂ ∂ ∂ ∂ . ∂2 . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . (1. deﬁned by ∂2 − ∇2 .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. . = −24. (1. not aﬀected by Lorentz transformations.33) (1. for instance.36) Exercises Exercise 1. (1. starting from ǫ0123 = 1.28) ν ν Note that gµ with one upper and lower index.∇ . ′ ′ . explicitly written as (p0 . (1.

but only appropriate combinations. 0) ˆ n2 = (0. One does not need to know . ˆ ˆ ˆ ˆ These are n1 = (0. a2 ] or [a− . 0) ˆ n3 = (0. 1.3. 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− .13 to give its actual value in kg. n1 . 1. n1 . n2 . a2 ] we can ﬁnd basis vectors n− . re = e2 /4πǫ0 me c2 to the Compton wavelength. 2. 1. ν. a2 . (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). g−− . g+− and g−+ . How are a+ and a− related to a+ and a− . 1. ]. [Note: what is the MKS unit for V/T?] Exercise 1. . Exercise 1.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a1 . ǫ0 . ρ. ˆ ˆ .71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). Also for the coordinates [a− . n2 . Also construct and calculate the Planck length Lpl . . a1 .511 MeV. 0. a1 . by a simple few-line reasoning [For instance: If {µ. (b) The coordinates (a0 . This demonstrates how a careful use of units can save a lot of work. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . 0. 0.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). e.(A · B) C using the properties of the tensor ǫijk given in section 1.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . a+ . 0. a+ . 1) ˆ . n3 . which is the Compton wavelength for the Planck mass. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. 0. 0. 0) ˆ n0 = (1. 3} the index α can only be equal to one of the indices in ǫµνρσ . a+ . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. α and me c2 = 0.2 Prove the identity A × (B × C) = (A · C) B . Exercise 1. me . a3 ) are the expansion coeﬃcients using the basis vectors n0 . n+ . (d) Use the value of the gravitational constant GN / c5 = 6. Compare it with the proton mass and use Eq. . c. σ} is a permutation of {0.

2) Acting on the wave function one ﬁnds for a free particle. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. p).3) Equations 2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions.7) k 0 = ± k2 + M 2 = ±Ek .5) with (k 0 )2 = k2 + M 2 . (2. i ∂ ∇2 ψ(r. t).1 and 2. But the replacement 2.4) where M is the particle mass. p2 = pµ pµ = E 2 − p2 = M 2 . The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. ∂t p −→ pop = −i∇.Chapter 2 Relativistic wave equations 2.3 are not covariant. 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. (2. φk (r. E = p2 . (2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). namely plane waves characterized by a wave number k. ∂t 2M (2. t) = 0. 2M (2.2. t) = ∂2 − ∇2 + M 2 φ(r. 2 + M 2 φ(r. t) = − ψ(r. one obtains (2.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ .1 The Klein-Gordon equation In this chapter. Although it is straightforward to ﬁnd the solutions of this equation. ∂t2 (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. t) = exp(−i k 0 t + i k · r). 8 .

(2. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. Probability is not positive: in quantum mechanics one has the probability and probability current ρ j = = ψ∗ ψ ↔ i i − (ψ ∗ ∇ψ − (∇ψ ∗ )ψ) ≡ − ψ ∗ ∇ ψ. ↔ ↔ (2. The KG equation.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. (2. rather than as a wave function. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory.5).e.2) is j µ = i φ∗ ∂ µ φ or (ρ. −k) . At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. 2. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2.14) ˜ ˜ Introducing φ(Ek . just as the dependence on time was in ordinary quantum mechanics. there are solutions with negative energy. ∂µ j µ = 0. we want the most general solution. The appropriate current corresponding to the KG equation (see Excercise 2. 2M 2M (2.3). having the KG equation as a space-time symmetric (classical) equation. relativistically the density is not a scalar quantity. there are no longer fundamental objections to mix up space and time. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. which is what Lorentz transformations do. j) = ↔ i φ∗ ∂0 φ.9) They satisfy the continuity equation. Then. j) of the four-current j. k) + ei k·x φ(−Ek .15) . −i φ∗ ∇ φ . (2.10) ∂t which follows directly from the Schr¨dinger equation.Relativistic wave equations 9 i. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. but rather the zero component of a four vector. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek .11) Therefore. This continuity equation can be written down o covariantly using the components (ρ. for which we need the interpretation of φ itself as an operator. 2. As we will see later both problems are related and have to do with the existence of particles and antiparticles. k) ≡ a(k) and φ(−Ek . And. This leads to the existence of negative densities. however. (2π)3 2Ek (2.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x).8) (2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. ∂ρ = −∇ · j.13) ˜ with (in principle complex) coeﬃcients φ(k).

−x) ≡ (˜µ ). x ∂µ ∂ µ + M 2 φ(˜) = 0. Another symmetry is found by complex conjugating the KG equation. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. (2. i. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. x) → (t.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes).e. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).23) .18) (2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. after restricting the energies to be positive). (2π)3 2Ek (2. Performing some Lorentz transformation x → x′ = Λx. which implies (xµ ) = (t. for which we consider space inversion.17) (2.6 We will explicitly discuss the example of a discrete symmetry. (2.15 one has elimated k 0 and in both equations k · x = Ek t − k · x.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) . φP (x) ≡ φ(˜) (P for parity). 2. (2π)3 2Ek (2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . Since a · b = a · ˜ it is easy to see that ˜ b. changing the sign of the spatial coordinates. This shows how parity transforms k-modes into −k modes.Relativistic wave equations 10 In Eqs 2.3 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. 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). (2.16) The consequence of this is discussed in Exercise 2. (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) .14 and 2.

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

(b) Show that one has current conservation for the current 2 We denote the commutator as [A. 2.2) in terms of the a(k) and b(k) using the expansion in Eq. αj } = 2 δ ij I. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. β} = 0.1) for a complex scalar ﬁeld current (exercise 2. and β 2 = I. B] = AB − BA and the anticommutator as {A. {αi .5 (a) As an introduction to parts (b) and (c).7 To solve the problem with positive and negative energies and get a positive deﬁnite density. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. 1 2 (b) Express now the full charge QV (t) (exercise 2. Note that a{µ bν} indicates symmetrization. i (a) Show that relativistic invariance. j 0 = ψ†ψ and j = ψ † αψ. Exercise 2. d3 x (∂ {0 φ)∗ (∂ i} φ). Is the time dependence of the result as expected? Note that the mode amplitudes will become creation and annihilation operators after quantization of the ﬁelds (Chapter 7). B} = AB + BA. t) = (−i α · ∇ + m β) ψ(r. We will encounter these quantities later as energy and momentum. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). 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 . 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 . Exercise 2.6 Write down the mode expansion for the Lorentz transformed scalar ﬁeld φ′ (x) and show that it implies that the Lorentz transformed modes satisfy a′ (Λk) = a(k).15. a{µ bν} ≡ aµ bν + aν bµ . 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). (c) Similarly. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . in terms of the a(k) and b(k).Relativistic wave equations 12 Exercise 2. . ∂t where ψ is a wave function and the nature of α and β is left open for the moment. also using the result in (a). t). ∂ ψ(r.e. i.

in which the parameter space forms locally a Euclidean space.g. the symmetry group describing rotations is embedded in the Lorentz group. n) where n is a unit vector and ˆ ˆ ϕ is the rotation angle. 0 ≤ ϕ ≤ π. The parameter-space of SO(3) is a ˆ ˆ ˆ sphere with radius π. 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. in particular each component of these spinors will satisfy this equation. 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. z )V . n) ≡ R(π. Any rotation can be uniquely written as R(ϕ. In quantum mechanics spin is described by a vector.1) for a rotation around the z-axis or shorthand V ′ = R(ϕ. 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. 3.3) ϕ=0 13 .g. e. In particular.Chapter 3 Groups and their representations The simple systems that we want to describe in a relativistically invariant way are free particles with spin. it has 3 components that we know the behavior of under rotations. and we must study the representations of the Lorentz group.e. z ) = 1 + i δϕ Lz with as generator ˆ 1 ∂R(ϕ.2) (3. however. −n). z ) Lz = i ∂ϕ 0 = i 0 −i 0 0 0 . For an inﬁnitesimal rotation around the z-axis one has ˆ R(δϕ.1 The rotation group and SU (2) The rotation groups SO(3) and SU (2) are examples of Lie groups. i. we want to investigate if there exist objects other than just a scalar (real or complex) ﬁeld φ. In a relativistic theory. electrons. Locally this parameter-space is 3-dimensional and correspondingly one has three generators. The KG equation will actually remain valid. provided we identify the antipodes. R(π. Since the KG equation expresses just the relativistic relation between energy and momentum. In this section we will investigate the requirements imposed by Poincar´ e invariance. Particles with spin then will be described by certain spinors. we also want it to hold for particles with spin. e. 0 0 (3. that is groups characterized by a ﬁnite number of real parameters. i.e.

Lj ] = i ǫijk Lk . thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. 0 (3.and y-axes 0 0 0 0 0 0 0 −i . x]] + [z. namely that there exists a bilinear product [. One way of viewing the parameter space. 2 (3. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx .5) Rotations in general do not commute. y ∈ A ⇒ [x. with real a’s and 3 (ai )2 = 1. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. x] = 0 (thus [x. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. z]] + [y. [x. . n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . These generators form a threedimensional Lie algebra SO(3).Groups and their representations 14 (n. They satisfy the commutation relations [Li . (3.7) = a0 0 −1 i 0 1 0 0 1 • [x. 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. ˆ therefore.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). [y. In the same way we can consider rotations around the x. Ly and Lz . which reﬂects itself in the noncommutation of the generators. Next. [z.4) R . and thus (σ · n)2 = 1. • [x. Locally this is a 3-dimensional Euclidean space and SU (2). for rotations around z for instance. x]). Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . z ) = lim N →∞ that are generated by i 0 . These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. (3. is a 3-dimensional Lie-group. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. ] that satisﬁes • ∀ x. ˆ R(ϕ. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ.1: the parameter spaces of SO(3) (left) and SU (2) (right). y]] = 0 (Jacobi identity).6) Summarizing. y] = −[y.2 π) (-n. y] ∈ A. π) (n.z N N or (Lk )ij = −i ǫijk .

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

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

For SO(3).g. If rotations leave H invariant.3). m = Dm′ m (ϕ. If a transformation U acts on χ. Jz = i 0 i 0 −i 0 0 0 −i 0 0 0 0 0 . 1 |1. to be precise the states ǫχ transform via −σ ∗ . For an element in the group G the corresponding point in the parameter space can be reached in two ways.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 . j. J− = 2 0 0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0 . e. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2. 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. however. however. χ) = eim ϕ dm′ m (θ) e−imχ . The group SO(3). There exist two diﬀerent types of paths. Consider the j = 1/2 representation of SU(2). 2 From the (hermitean) representation Φ(g) of G. . (j) (3.19) (3. one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . 2 |1. if χ → σχ. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. the possibility thus exist that some extensions will not be well-deﬁned representations. one can look at the conjugate representation. m′ |UE (ϕ. For those global representations. ǫy . Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states. SU (2) is the covering group of SO(3). the extension from a local one must be unique. ǫz } are 0 0 0 0 0 i 0 Jx = 0 0 −i . Of all groups of which the Lie algebras are homeomorphic. the topological structure of the group is important. −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . the simply connected group is called the (universal) covering group. is not simply connected.e. Jy = 0 0 0 . 0 ≡ ǫ0 ≡ ǫz . the conjugate transformation U ∗ acts on χ∗ . however. i. that is any closed curve in the parameter space can be contracted to a point. θ. Given a representation. θ. For SU(2). J = 0 0 + 0 0 0 0 0 0 0 0 √ 2 0 √ 0 . For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles. This turns out to be the case for all half-integer representations of the Lie algebra. For a decent (welldeﬁned) global representation. or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . contractable and paths that run from a point at the surface to its antipode. Explicitly. θ. χ)|j. dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m . This works for SU (2).20) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. χ) ≡ e−iϕJz e−iθJy e−iχJz . all states in the representation space have the same energy. 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. If the group is simply connected. the conjugate representation is not a new one. UE (ϕ. (j) (j) ′ The spin matrices for that Cartesian basis {ǫx . however. For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1. then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ.

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. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .25) (3.27) (3. σ/2 in a two-dimensional (spin 1/2) case.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . (ii) |Λ00 | ≥ 1.22) (Note that xT is a row vector). G = 10 (3.Groups and their representations 18 3. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.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. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. it is convenient to use a vector notation for the points in Minkowski space.26) (3. (det Λ = +1 is called proper. (3. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.g. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . Writing x as a column vector and the metric tensor in matrix form. consisting of the Lorentz group and translations.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. e To derive some of the properties of the Lorentz group. 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). In this section we consider the Poincar´ group.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. e. We considered the generators and looked for representations in ﬁnite dimensional spaces. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. From this property.

inﬁnitesimally given by ΛB (η. z ) ≈ ˆ ˆ I − i η K 3 . z ) ≈ I + i ϕ J 3 . In L↑ . K j ] = −i ǫijk J k . J 3 ). z ) = exp(i ϕ J 3 ).32) 0 − 0 −1 0 0 0 0 −1 . The commutator of two boosts in diﬀerent directions (e. since the generators K do not form a closed algebra. Rotations around the z-axis are given by ΛR (ϕ. K j ] = i ǫijk K k . 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). Returning to the global group. J j ] = i ǫijk J k . J = (J 1 . J 2 . one distinguishes rotations 0 + and boosts. we have found the generators of rotations. From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. i proof: use ΛT GΛ = G and ΛGΛT = G. 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 . βγ = sinh η. K = (K 1 . inﬁnitesimally given by ˆ ΛR (ϕ. This is the origin of the Thomas precession. since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η.Groups and their representations = i=1 (Λ0 )2 .28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. 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. 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. [J i . Using these explicit transformations. z ) = exp(−i ηK 3 ). K 2 . which satisfy the commutation relations (check!) [J i . it is easy to ﬁnd the following typical examples from each of the four classes.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3.30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3. (3. and those of the boosts.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↑ . 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 . (3.g. [K i . K 3 ).

Also this group can be divided into four parts. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations.36) (3.39) (e. [M µν . There are six generators. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) .38) Explicitly.g. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . This is a + normal subgroup and the factor group is the Vierer group.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 .g. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). inf (3. K = M ). The extension to the Poincar´ group is e ↑ straightforward.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . We will just require Poincar´ invariance for the generators themselves. (3. i i0 i i0 1 ijk M jk 2ǫ (3. ǫ). we continue using covariance arguments. (3. a). In order to ﬁnd the commutation relations including the translation generators.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. 3 30 with η = ω and K = M (e. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. Of the four parts only L↑ forms a group. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . 3. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. inﬁnitesimally approximated by (I + ω. P+ etc. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .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) (3.33) These four transformations form the Vierer group (group of Klein). Summarizing.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. We therefore ﬁnd (again) that there are six generators for the Lorentz group.

P j ] = i δ ij H/c2 . P ρ ] = 0. Inﬁnitesimally. a) = Λµν P ν or U † (Λ. [J i . The eigenvalues of these will be the four-momentum pµ of the state. [P µ . (3. . [J i . P µ |p. ǫ) = 1 − ωαβ M αβ + i ǫα P α . Explicitly. . P ) in terms of the Hamiltonian and the three-momentum operators.43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . H] = [J i . [K i . P ν ] [M µν .46) [K i .41) 2 At this point. ν (3. P j ] = i ǫijk P k . . writing the generator P = (H/c. the generators M αβ no longer exclusively act in Minkowski space. H] = i P i . other states in that representation are obtained by the action of operators outside the maximal commuting set.45) Note that the commutation relations among the generators M µν in Eq. one obtains 2 [P i . For instance.44) (3. Taking any one of the states in an irreducible representation. a)P µ U (Λ. 3. as part of the set. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . check that U1 U2 corresponds with Λ1 Λ2 ).5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. . known from non-relativistic quantum mechanics are obtained. J j ] = i ǫijk J k . a) = Λ−1 µ P ν . They just state that M µν transforms as a tensor with two Lorentz indices. . 2 2 (3. K j ] = −i ǫijk J k /c2 . that commute e among themselves. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. the generators J 2 and J 3 in the case of the rotation group. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. (3. = −i (g µρ P ν − g νρ P µ ) . [J i .47) (3.Groups and their representations 21 inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. K j ] = i ǫijk K k . . P j ] = [P i . .42) (to decide on where the inverse is. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . (3. 3. Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . (3. We have here reinstated c. For instance. The generators of the translations are the (momentum) operators P µ . leads to U (Λ. = pµ |p. These form a group called the little group associated with that four vector. [K i . a)P µ U † (Λ. H] = 0.38 could have been obtained in the same way. the generators J± in the case of the rotation group.48) . (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation).

2 (3.51) The following properties follow from the fact that W µ is a four-vector by construction.52) (3. Pν ] = [Wµ . i ǫµνρσ W ρ P σ . .54) This will enable us to pick a suitable commuting ’spin’ operator. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . = = i exp − ǫµνρσ M µν pρ sσ |p. (3. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. (3. Wν ] = −i (gµα Wν − gνα Wµ ) . Wα ] = [Wµ . Massive particles: p2 = M 2 > 0.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 . . which can thus be chosen spacelike. p0 > 0. . while the others have no physical signiﬁcance.49) (3. 0. . .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . the third case represents the vacuum. 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. M ǫijk . 0. From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.56) . . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).53) (3. 2 exp (−isµ W µ ) |p.55) commute with all generators and therefore are invariants under Poincar´ transformations. . .

e. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). . it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 0. M2 (3. n1 (p).59) (3. Together with the four momentum states thus can be labeled as |M. W j (p)] = i M ǫijk W k (p). 0).64) such that in an arbitrary frame where the four vectors p. 0). i (3.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . s. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). show that W /M can be interpreted as the intrinsic spin. ms . 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. (0. In that case the vectors ni (p) are just the space directions and W = (0.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). 0. 2 .. |p3 |).65) . 1. (3. 0. W i (p) = −W · ni (p)). (0. 0. 0. The commutation relations [W i (p). n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. 1.61) one sees that S i (p) i =− W2 . 1 2 (3. M J) with components W i = (M/2)ǫijk M jk . Having made a covariant decomposition.57) (i.58) i. 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 . (1.e. . p0 > 0. (3.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. M2 (3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. . 0). p. 3. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 1.57 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. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 0.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).

Thus 2 λ(p) = J ·p . 3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises).67 has a vanishing right hand side). W 2 (p)] [λ(p). thus.13. |p. W 2 (p)] = = = i M 2 λ(p).1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations.Groups and their representations The above does actually include the massive case. A massless particle. (3. But if M = 0 one has a diﬀerent algebra (Eq. Note that angular momentum does not contribute to helicity as L · p = 0.68) (3.73) (3.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). W 2 (3. which can take any integer or half-integer value. 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). This state is in physical applications nondegenerate and is denoted by |0 . σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). W 1 (p)] [λ(p). however. Thus we have P 2 . −i W 1 (p). implying continuous values for the spin (actually for W i = M S i ). U (Λ. 3. We don’t know of any physical relevance. The vacuum: pµ = 0. P µ . (3. The algebra of W 1 .70) as a commuting set with zero eigenvalues for P 2 and W 2 . i W 2 (p). Wν ]. and gives [W 1 (p). 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). λ . |p| (3. W 2 and λ is derived from the general commutation relations for [Wµ . 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 . is characterized by its momentum and the helicity. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). W µ (p) = λ(p) P µ . 1 2 24 (3. a)|0 = |0 .71) which is called the helicity.67) (3. The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with W 0 (p) = − 1 ǫ0ijk Mij pk = J · p. Exercises Exercise 3.

Excercise 3.4 (a) If the states in a Hilbert space (unitarily) transform as φ′ = U φ = eiJ φ and O is an operator acting on φ in that Hilbert space. Tr(σi σj σk ) = 2i ǫijk . . In both of these cases the symmetry is broken by the charges ¯¯ of the particles. Examples are the quark (u.Groups and their representations 25 for any vectors a and b. (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). (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. Note the diﬀerence with the (usual) spin 0 and spin 1 states composed from two spin 1/2 states. respectively). Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and thus Tr(σi σj ) = 2 δij . Exercise 3. the operators Jk must be hermitean. . (c) In situations in which going to the conjugate representation (which is used to describe antiparticles) cannot be followed by a redeﬁnition of the basis. u). show that one matrix ǫ exist such that J c = ǫ J ǫ† . for instance. one has to explicitly include this rotation. d) doublet representation for SU (2)isospin or the Higgs (φ+ .e. Show that the Lie-algebra representations J and J c are equivalent. . Both particle and antiparticle have ’spin 1/2’. Is this easy to understand. Writing A∗ = exp(i α · J c ). This is true for SU (2) but it is. Exercise 3. triplet and anti-triplet representation. the pions and the isospin 0 meson state. Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) . and calculate U (a) p U −1 (a) . (b) This establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle.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. what are the matrices J c in the Lie-algebra representation describing the spin rotations for two-component ’anti-spinors’. φ0 ) doublet representation for SU (2)weak .2 Show that for a unitary operator U = exp[iαk Jk ] with real coeﬃcients αk . show that the transformation for the operators is given by O′ = U O U −1 = O + i [J. i. Now write down the three isospin 1 meson quark-antiquark (meson) states. Show that if det(U ) = 1. the (non-strange) η-meson state. Show actually that ǫ = 2 eiπJ (up to a phase). O] + . The appropriate isospin doublet to be used for antiquarks then is (−d. For a spin 1/2 antiparticle the representation A∗ is needed. not true for SU (3) (ﬂavor or color properties of particles know two distinct deﬁning 3-dimensional representation. the trace of the operator Jk is zero (Convince yourself that det(eA ) = eTr A by diagonalizing A.

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

Chapter 4 The Dirac equation 4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. 2 (4.1) (4. The group SL(2. The matrices in SL(2. becoming the deﬁning representations of SL(2. They will be labeled by two angular momenta corresponding to the A and B groups. 0) Type II : (0. i ǫijk B k .3) (4. respectively. It is simply connected and forms the covering group of L↑ . With this choice of parameter-spaces the plus and minus signs are actually relevant. (4. Special cases are the following representations: Type I : (j.7) From the considerations above. j) K = −i J (B = 0). (j. B j ] = i ǫijk Ak . (4. 2 2 σ σ J = . C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . K = −i . Aj ] = [B i . 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 .8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. This tells us how to ﬁnd the representations of the group.1 The Lorentz group and SL(2. K = i J (A = 0).4) (4. K = +i .9) (4. 0. j ′ ).10) . M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) .2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2).5) = 1 (J + i K).6) (4. similarly as the homeomorphism between SO(3) and SU (2). C) is the group of complex 2 × 2 matrices with determinant one. C). They precisely correspond to the two types of representations that we have seen before. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. 2 1 (J − i K). 2 2 (4. B j ] = [Ai .

With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. This shows that M ∗ ≃ M .15) (4. H(η) = exp(−η · σ/2). η → Hη. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . (4. One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . The following relations for + rotations and boosts are useful. C). ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. H ≡ (H † )−1 = H −1 ). if there does not exist a unitary matrix S. 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).The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. ξ −→ Hξ. thus. (4.13) 2M (E + M ) (exercise 4.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. (4. i. σ µ ≡ (1. C). In fact. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H.18) As already discussed for SU (2). −σ). C) or L↑ . σ). 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. U σ µ aµ U −1 = σ µ ΛR aµ . (ǫ∗ = ǫ.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ).e. however. ˜ (4.19) . The Lorentz transformations Λ(M ) and Λ(M ). U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4.11) Considering ξ and η as spin states in the rest-frame. ˜ = σ µ Λ B aµ . 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. (4. cannot be connected by a Lorentz transformation belonging to L↑ . Eqs 4. such that M = SM S −1 . ǫ−1 = ǫ† = ǫT = −ǫ). C) and. while ±ǫη ∗ transforms as a type-I + spinor.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.12) ξ → U ξ. namely H(η) = exp(η · σ/2). one can use a boost to the frame with momentum ˆ ˆ p. Λ(ǫ) ∈ L+ .1). (4. This is not true for the two-component spinors in SL(2. η → U η. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . thus.16) = σ µ Λ R aµ . there exists the matrix ǫ = iσ 2 .

1. = (4. m) = χm . i. Within the Lorentz group. The representing member of the class P− is the parity or space inversion operator Is . but rather the operators W i (p) should be used. The behavior is the same for classical quantities. (axial vector). 4. 1 ) and ( 1 . −→ −r (vector). Thus M ∗ and M are not equivalent within SL(2. ) −→ ( . C).22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq.17. m) = η(0. (pseudoscalar). 2 ) and ( 2 . For the spin 1/2 representations of the Poincar´ group including parity we. (4. For a particle at rest only angular momentum is important and we can choose ξ(0. A parity transformation changes aµ into aµ . the aij elements of a tensor will not change sign. they cannot be connected by a unitary 2 2 transformation. 4. This provides the easiest way of seeing ˜ what is happening.e. −→ M (Λ) (4.e. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). m) = χm uL H(p) 0 . but to L↑ .g.13. etc. we obtain for the two components of u which we will refer to as chiral right and chiral left components.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0.The Dirac equation 29 where The matrix I2 does not belong to L↑ . ǫµνρσ will change sign. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . −→ −p (vector). It has the same eﬀect as lowering the indices. but by a transformation belonging ↑ ↑ to the class P− . the well-known two-component spinor for a spin 1/2 particle. We also have seen that the representations (0. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. e. 0) are inequivalent.20) 4. (vector). From the deﬁnition of the representations (0. The i angular momentum operators J are represented by σ i /2. Taking M (Λ) = H(p). 0 χm uR H(p) . 2 2 (4. the boost in Eq. (scalar). 0). that have the correct commutation relations. they can be connected. therefore. generators. 1 ) and ( 1 . −a).23) u(p. must combine the repe resentations. a) and ˜ a = (a0 . as we have seen in section 4. C) are suitable for representing spin 1/2 particles. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . where a = (a0 .21) In nature parity turns (often) out to be a good quantum number for elementary particle states. Although these operators cannot be used to label the representations. 0) of SL(2. we have seen that in the rest frame W i (p)/M → J i . i. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle.

25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. γ } ∂µ ∂ν + M 2 ψ(x) = 0. 2 (4. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). which in general is a linear equation in pµ .30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). (4. (4. (4. We will discuss another explicit realization of this algebra in the next section. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.32) suppressing on the RHS the identity matrix in Dirac space. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. this can be rewritten 1 µ ν {γ . It is of a form that we also played with in Exercise 2.7. . From this one obtains the Cliﬀord algebra for the Dirac matrices.29) which is an explicit realization of the (momentum space) Dirac equation. (4. (4. p (4. = H −1 (p).33) deﬁne for a four vector a the contraction a = aµ γµ / . 0 H 2 (p) uR uR . The ﬁrst indeed is solved.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies.24) (4. γ ν } = 2 g µν . This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.27 is known as the Weyl representation. The general requirements for the γ matrices are thus easily obtained. 4. {γ µ . The explicit realization appearing in Eq. (iγ µ ∂µ − M ) ψ(x) = 0. = (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 .28) which is a covariant (linear) ﬁrst order diﬀerential equation. Applying (iγ µ ∂µ + M ) from the left gives (4.26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0. Since ∂µ ∂ν is symmetric. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).

From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).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 .35) is indeed positive and j 0 can serve as a probability density. 4. γ ν } = 2 g µν . (4. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. (4. This problem was solved by Dirac through the introduction of a negative energy sea.g. This can most easily be seen in the particle rest frame. 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.37) Using the explicit form of γ in the Weyl representation (see Eq. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. The second problem encountered before. (4. The Dirac sea forms the vacuum. Relying on the Pauli exclusion principle for spin 1/2 particles. opposite charge). The density j 0 = ψγ 0 ψ = ψ † ψ (4.3 General representations of γ matrices and Dirac spinors {γ µ . From the vacuum a particle can be removed.36) (4.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. This hole forms an antiparticle with the same mass. remains. but with properties such that it can be annihilated by the particle (e. E = +M (twice) and E = −M (twice).1: The Dirac sea of negative energy states and antiparticles. the one of the negative energy states.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices.27). . where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.34) for the adjoint spinor ψ ≡ ψ † γ0 . 4. one sees that there are two positive and two negative eigenvalues. .

= S(γµ )W. which in the case of the Weyl representation are just the upper and lower components.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. the rotation and boost generators in Weyl representation in Eqs 4. is 1 1 1 (4.41) (4. =ρ ⊗1= 0 −1 3 (4.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .43) S=√ . S −1 .R. For instance in the Weyl representation (but valid generally).R. γ 0 ]. 4! (4.40) (4.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . −→ Sψ.47) (4.3) . it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . while L−1 γ µ L = Λµ γ ν . 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 . Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.46) (4. −→ Sγµ S −1 . For example.48) are projection operators.45) (γ5 )W. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . 0 (4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .R.R. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. γ σ ]. 2 (4. (γ5 )S.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. (4. We note that ψ → Lψ and ψ → ψ L−1 . K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.9 and 4. (γµ )S. that project out chiral right/left states. γ 2 ]. γ µ } = 0 and explicitly one has 0 1 1 .

Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . x = (t. One has ψ c = −ψ T C −1 . ∂ with the solution ψ c (x) = ηc Cψ (x). e. (C)S. ψ= (4. In any representation a matrix C exist.57) (4. The latter implies that the conjugate of a right-handed spinor. = = 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. reverse the charge of the particle.53) (4. (C)W. γ5 ] = 0. such that T Cγµ C −1 = −γµ . Explicitly.) and all representations connected via a real (up to an overall phase) matrix S.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. 0 −ǫ 0 0 −ǫ 0 . ∂ (4.R. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . that brings ψ → ψ c . for instance.R.52) η ξ one can apply space inversion. In S.59) −→ ψ c = C ψ = η ǫ ξ∗ . called charge conjugation. x) → x = (t.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. C is real and one has C −1 = C † = C T = −C c and [C. Some properties of C are C −1 = C † and C T = −C. but it does. which is again a solution of the Dirac equation. is a left-handed spinor. ∂ (4.56) (i/ − M ) ψ c (x) = 0. This symmetry does not change the spin 1/2 nature.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. 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. ψ= (4. ψR .54) e. T (4.58) where ηc is an arbitrary (unobservable) phase. (i/ − M ) ψ(x) = 0. T (4. parity. −iσ 2 0 −ǫ = . As with parity we look for a transformation.g.6).R. (or W.55) (4.g. usually to be taken unity.R. = 0 ǫ iσ 2 (4. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.

the spinors satisfy C uT (p. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. ∂ −iσ · ∇ −i ∂t − M (4. s)v(p. s′ ) = 2M δss′ . s′ ) = 2Ep δss′ . v (p. s′ ) = −2M δss′ . s) v = . s) and C v T (p.68) (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. s).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)u(p. v(p. s)¯(p. ψ(x) = u±s (p) e−i p·x . s ) = v † (p. p (4. s) = Ep + M σ ·p Ep +M χs .4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. s ) = v (p. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s) = u(p. ¯ ¯ † ′ u (p.64) (4. (2π)3 2Ek (4. the spin polarization vector in the rest frame is the starting ˆ point. After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) = s d3 k u(k. s) = v(p. In an arbitrary frame one has s · p = 0 and s(p) can e.70) = Ps = ˆ 0 1∓σ·s 2 2 .65) (4. ψ(x) = v ±s (p) ei p·x . s). ⇔ (/ + M ) v(p) = 0. where u satisﬁes −σ · p Ep − M u(p) = 0.63) u(p. In that frame is a spacelike unit vector sµ = (0. be obtained by a Lorentz transformation. ¯ (4. s) u = . 2M 2M (4. s)u(p. s) ei k·x d∗ (k.1). s) = u(p. s)v(p. (4. with E = Ep = + p2 + M 2 .66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. the best representation to describe particles at rest is the standard representation. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors.62) where χs are two independent (s = ±) two-component spinors.60) ψ(x) = 0. The ¯ ¯ solutions are normalized to u(p. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. s) .61) σ·p −(Ep + M ) There are also two negative energy solutions. s ) = 0..g. s) + v(k.69) = − s In order to project out spin states. ¯ ′ ′ Explicit solutions in the standard representation are χs . s)¯(p. ⇔ (/ − M ) u(p) = 0. s)v(p. s)u(p. s) e−i k·x b(k. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. 2M 2M −/ + M p v(p.

u(p.74) . The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . In principle massless fermions can be described by twocomponent spinors. v(p. λ = −1/2) = u(p. chirality. For instance with the z-axis as spin quantization axis. +1/2) = √ 0 E +M E+M . (4. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. λ) = u(p. λ) and C uT (p. 2 2 0 √ 2 √ 2 − E+M − E−M 0 . γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. λ). +1/2) = √ u(p.73) Also for helicity states C uT (p. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). Note that for solutions of the massless Dirac equation /ψ = 0.48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). while the negative helicity (λ = −1/2) is in essence lefthanded. = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. The chirality projection operators in Eq. √ 0 u(p.71) p − i p2 . one has in standard representation: 0 E+M E+M 1 1 0 . 4.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 ). helicity states. −p3 −p1 − i p2 1 v(p. (4. 1 u(p.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . . This means that in the solution space for massless fermions the chirality states. −1/2) = √ (4. −1/2) = √ E + M −(E + M ) 0 . λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . λ = − 1 ) = √ √ v(p. Therefore. λ) = v(p. λ = + 1 ) = √ + E +M − E −M . One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . λ = +1/2) = v(p. λ = −1/2) = √ − E+M √ 0 0 E+M (4. etc. Explicit examples of spinors are useful to illustrate spin eigenstates. λ = − 1 ) = √ E + M − E − M . u(p. p ψR/L ≡ PR/L ψ are independent solutions. v(p.

s′ )) . s)(γ ν )kl u u s. s)(γ ν )kl ul (k ′ . s′ )(γ µ )ij uj (k. k k 2 . s)γ µ u(k ′ .) In order to show the use of the above relations. (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ . (5) γ µ γ λ γ ν γ ρ γ σ γµ = 2 γ σ γ λ γ ν γ ρ + γ ρ γ ν γ λ γ σ . s′ )γ µ u(k. For this we ﬁrst have to prove (do this) that (¯(k. however. s′ )(γ µ )ij uj (k. ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ . s)γ µ u(k ′ . s′ )¯i (k ′ . (8) σµν ≡ (i/2)[γµ . (Use (γ5 )2 = 1 and pull one γ5 through.s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . (10) γ µ γ ν = g µν − i σ µν (11) γ µ γ ν γ ρ = S µνρσ γσ + i ǫµνρσ γσ γ5 with S µνρσ = (g µν g ρσ . ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . s)γ µ u(k ′ . (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ). (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . u u (4. s′ )) = u(k ′ . k ′ ) = 2 ′ s. (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. It is. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. 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. (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. more convenient ¯ to use the projection operators introduced earlier.The Dirac equation 36 4. etc. namely the calculation of the quantity 1 ∗ (¯(k. s′ )) (¯(k.76) ul (k ′ . In principle one can take a representation and just calculate the quantity u(k. s)¯k (k. s′ ). (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. s)¯k (k.) (2) Tr(1) = 4.s′ ∗ (4.75) Lµν (k.g µρ g νσ + g µσ g νρ ). we will give one example. a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b.s which appears in quantum electrodynamics calculations (see Chapter 10) for the emission of a photon from an electron changing its momentum from k to k ′ (k 2 = k ′2 = m2 ). s′ ) ¯ u s.g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. s)γ ν u(k ′ . γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ .5 γ gymnastics and applications An overview of properties can be found in many text books. s). (6) Tr(γ5 ) = 0.

(4. by letting them act on the four basis spinors u(p.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜).1 Prove Eq.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. 4 2 Exercise 4. s) u p /+M = 2M 2M both sides are projection operators. Do this e. 0 if n is odd. H(p) = exp η·σ 2 = M +E+σ·p . s) and v(p. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . of which only the traces of four and two gamma-matrices are nonzero. a 4 a · b. ˜ Exercise 4. s)¯(p. Excercise 4. where x = (t. Exercise 4. −x) is also a solution of the Dirac equation. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . Show that the equality holds.77) Exercises Excercise 4.g. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.4 Show that in P+ ≡ s=±1/2 u(p.5 Show starting from the relation {γµ . x → x. x ˜ . 1 i −i S ρσ = − [γ ρ .6 Apply space-inversion.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. s). 4.13.g. γ σ ] = σ ρσ . e.

Calculate for now just |M |2 . (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. the decay would even be forbidden. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. The pion has spin 0. 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.e. me = 0. ν p Note: how such parametrizations work will be explained in chapter 8. Calculate the value of k for M R = 0. r). we want to conﬁne this solution to a spherical cavity with radius R. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. mµ = 105. there is no current ﬂowing through the surface of the cavity. (b) Next. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). . you need to determine the corresponding conﬁnement condition for ψ. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. We will not pursue this chiral freedom at this point and take α = 0.57 MeV. Our starting point is the transition amplitude given by M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ). which is the most important factor needed to ﬁnd the decay width (inverse lifetime).511 MeV). Do you understand why? The calculation of the life-time is completed in chapter 11.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). 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. is a stationary solution of the (free) Dirac equation.66 MeV.5 and M → ∞. i. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. 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.The Dirac equation 38 Exercise 4.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. n It turns out that there is actually a whole family of conditions that provide conﬁnement. Excercise 4. Note: To prove this. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0.8 × 10−6 . Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. M R = 1. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . namely i / ψ = eiαγ5 ψ.

with ∂µ V µ = 0 (Lorentz condition).4) λ=0.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. These are solutions of the Klein-Gordon equation with in addition a condition. there are then indeed three independent possibilities. M2 (5. λ) .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .± 5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .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. namely ǫµ (k0 ) = (0. A). µ (2π)3 2E µ (5. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. In an arbitrary frame ǫµ (k) with λ = 0. ±(1) are obtained by boosting the restframe vectors. In the restframe.7) .5) −E B2 . ǫλ ).Chapter 5 Vector ﬁelds and Maxwell equations 5. 39 (5.1) The Lorentz condition implies that k µ ǫµ (k) = 0.1 Fields for spin 1 In section 3. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. These can describe spin 1 states in the rest-frame. (2 + M 2 )Vµ = 0. just as the two-component Dirac spinors did for spin 1/2. (5. 0).2) (5. The vectorﬁeld is therefore (λ) suited to describe spin 1. where k = k0 = (M.6) where Aµ is the four-vector potential (5.

˜ x Note that via parity transformation one obtains another solution.15 the d’Alembertian is replaced by 2 + M 2 . only two independent vectors remain ǫ(±) µ (k) = (0. 0). Furthermore. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ .5. (5. It states the absence of magnetic charges and Faraday’s law. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. transverse or Coulomb gauge). AP (x) = Aµ (˜). The equation in vacuum. Therefore if k µ = (|k|.11) This equation is not aﬀected by a gauge transformation. (5.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. j) the Maxwell equations read ∂µ F µν = j ν . (5. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.8) (5.e. ∂µ Aµ = 0.15) (5. −i.16) (5. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0.12) (5. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). it will vanish at all times. 0. Although the Lorenz condition is a constraint. 0.13) Hence. (5. . |k|). the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. moree over. i.17) (radiation. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. For electrodynamics one has furthermore the freedom of gauge transformations. for M = 0. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). Under a gauge transformation Aµ −→ Aµ + ∂µ χ. in accordance with what we found in section 3. Taking the divergence of this equiation. (5.14) of which the solutions give the Li´nard-Wiechert potentials. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. 5. 2Aµ = 0. ∓1. 40 (5.10) the electric and magnetic ﬁelds are unchanged. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. if in Eq. if ∂µ Aµ = 0 for large times |t|. As remarked above. One possible choice is the gauge A0 = 0 (5. in which case one has 2Aµ = jµ .

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

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

∂t ∂B = 0.9 explicitly give the equations ∇ · E = ρ. ∂t ∇ · B = 0. .1 Show that Eqs. ∇×B =j+ ∇×E+ ∂E .Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5.8 and 5. 5.

thus t′ = t + δτ. As an example.8) 44 .2) The principle of minimal action looks for a stationary action under variations in the coordinates and time.3) (6. ˙ 2 (6. ∂x (6. x (t) = x(t) + δx(t). and the total change x′ (t′ ) = x(t) + ∆x(t).1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). (6. x) = K − V = ˙ 1 mx2 − V (x).6) The ﬁrst term leads to the Euler-Lagrange equations. These equations can be obtained from a lagrangian using the action principle.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.Chapter 6 Classical lagrangian ﬁeld theory 6. d dt ∂L ∂x ˙ = ∂L . t1 (6. ˙ p= ∂L . ∂x ˙ (6. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 .1) and as an example the lagrangian L(x.5) with ∆x(t) = δx(t) + x(t) δτ . recall classical mechanics with the action t2 S= t1 dt L(x. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. x) ˙ (6.

− d3 x .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). + ∂L ∆x − H δτ ∂x ˙ t2 . (R3 .6 can be rewritten as δS = . ∂µ φ(x). indicated with δF [φ]/δφ should pose no problems.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . ∂µ φ.11) λ=0 of which p and H are the simplest examples related to space and time translations. which in classical mechanics vanishes at the boundary) does not play a role. . L (φ(x). (6. = ∂R (R3 . . ∂µ φ(x)). their derivatives and possibly on the position.9) which is done because the ﬁrst term (multiplying ∆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 .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . i.t2 ) d3 x . We will come back to it in a bit more formal way in section 9. ˙ (6. with ∆φ(x) = δφ(x) + (∂µ φ)δx . Consider a lagrangian density L which depends on these functions. R (6.18) 1 Taking a functional derivative. x) ≡ p x − L . In general any continuous transformation. . x′ ) − d4 x L (φ. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 .2. . 6. x).Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. The hamiltonian H is deﬁned by H(p. Variations in the action can come from the coordinates or the ﬁelds. t1 (6. . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds.14) (6. . ˙ x The second term in Eq. δ(∂µ φ) (6. . ∂µ φ(x)) = R d4 x L (φ(x). think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ.12) Here R indicates a space-time volume bounded by (R3 . ∂µ φ′ . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ .13) (6. . t2 ). λ=0 (6. dσµ . .t1 ) (6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x).e. t1 ) and R3 . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). = φ(x) + δφ(x) (6. H is a conserved quantity. but which in 3 dimensions includes conserved quantities related to rotations and boosts.15) d4 x′ L (φ′ .

27) (6. (6. 2 ψ = 0. (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).28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration). the second term is irrelevant. One easily obtains (2 + M 2 )φ(x) = 0.21) which diﬀer only by surface terms. 2 (6.26) The Dirac ﬁeld The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ. ∂ (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. 2 ∗ (6.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.25) (6. leading to (2 + M 2 )φ(x) = 0. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ.2 Lagrangians for spin 0. (6. δL δL ∂µ = .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. leading to the Euler-Lagrange equations.24) (6.29) .23) (6. (2 + M )φ (x) = 0.19) δ(∂µ φ) δφ 6. The integrand of the ﬁrst term must vanish. Using the variations in ψ (in the symmetric form).20) (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found.

that the lagrangian separates into two independent parts for M = 0.32) There exists another possibility to write down a mass term with only one kind of ﬁelds. 1 M η † ǫη ∗ − M ∗ η T ǫη .31 is given by LM (Dirac) = −M ξ † η + η † ξ . it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left ﬁelds ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors.38) (6.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. (6.7). We note that of which the left part coincides . ψL ≡ (ψL )c = C ψL = 0 (6.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . or equivalently separate the ﬁeld into right. the mass term in the Dirac lagrangian 6. 2 αβ = −βα. Using the twospinors ξ and η.36) (6. but note the factor 1/2 as compared to the Dirac lagrangian. 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.31) showing e. The Majorana case is in fact more general. which comes because we in essence use ’real’ spinors.g.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.34) (6. Peshkin and Schroeder or Exercise 12. and thus (β ∗ α)∗ = α∗ β = −βα∗ . Vector ﬁeld From the lagrangian density L 1 1 λ 2 = − Fµν F µν + M 2 Vµ V µ − (∂µ V µ ) 4 2 2 1 µ = V (∂ 2 + M 2 )gµν − (1 − λ)∂µ ∂ν V ν . L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . ∂ ∂ 2 2 (6. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 .g. . namely2 LM (Majorana) = + (6. (6.and lefthanded ones. 2 (6. (6. The reasons for Grassmann variables will become clear in the next chapter. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).

which in a consistent picture precisely generate the symmetries. discussed in the next chapter.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x).18. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ.44) δL ∆φ + Θµν (x)δxν . one can consider the variations at the initial or ﬁnal surface. (6. δ(∂µ φ) (6. . ∂µ J µ (x) = 0. Q(t1 ) = Q(t2 ). we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . As an example for the classical ﬁeld. . (6. (6.47) ..45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .41) (6. these conserved quantities become operators.3 Symmetries and conserved (Noether) currents Next. In this case. (6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). 6. δ(∂µ φ) (6. requires the presence of a conserved quantity. δ(∂µ φ) For continuous transformations. Q= d3 x J 0 (x. t).42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. which is the space-integral over the zero-component of a conserved current. Returning to δS the surface term is rewritten to δS = R d4 x . consider U (1) transformations of the Dirac ﬁeld proportional to the charge e.. one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0.46) For quantum ﬁelds (φ operator!) things become more subtle as it it is not possible to specify ˙ for instance φ and φ on the surface σ1 . the presence of a symmetry that leaves the lagrangian invariant. we consider the second term in Eq.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . In particular when the lagrangian is invariant under symmetries.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν .40) 6.σ1 with σ = (R3 . These do not aﬀect the dynamics and will give rise to conserved currents. + ∂R dσµ dσµ ∂R ≡ where d x . (6. t).

˙ Q = d3 x j 0 (x.53) These are the energy and momentum. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). 1 (6. (6. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν .54) (6.51) (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . e. In the next section we will see the quantity show up as the charge operator. (6. including e the space-time translations and the Lorentz transformations. we obtain (since δxµ = 0).4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). t) is the conserved charge (Q = 0). (6.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν .48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. Summarizing.52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld. 6.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.g. From Noether’s theorem one sees that the current Θµν ǫν is conserved. H = d3 x Θ00 (x).50) (6. (6.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). .3) j µ = i e φ∗ ∂µ φ. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . 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. 2 (6. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. = 0.49) These currents are conserved as discussed already in chapter 2. The integral over the zero-component. ↔ (6.

In general this tensor is not symmetric.64) (6.70) . (6. (6.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. (6. φ(x).67) i. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). Any lagrangian containing derivatives. 2 (6. (6. independent of x. −i eΛ(x) ∗ i eΛ(x) (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).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 . ∂µ φ(x) + i e ∂µ Λ(x) e (6. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. however. e.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.69) i eΛ(x) φ (x).e. the angle of the transformation depends on the space-time point x. in which the U (1) transformation involves an angle e Λ. (6.63) µν = ∂µ Θ . is not invariant under local gauge transformations. ∂µ T µν (6.61) A ﬁnal note concernes the symmetry properties of Θµν .g. one has a tensor that satisﬁes (exercise 6.3). 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 . Deﬁning T µν = Θµν − ∂ρ Gρµν .3 we have seen global transformations or gauge transformations of the ﬁrst kind. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).65) M µρσ = T µρ xσ − T µσ xρ . where H ρµν is the quantity appearing in M ρµν .68) (6.66) In section 6.58) (6.6) T µν = T νµ . • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .59) (6. 6. • Vector ﬁeld Aµ : −i Sρσ = aρσ .

1 L (φ. (6.78) . The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ.76) We note ﬁrst of all that the coupling of the Dirac ﬁeld (electron) to the vector ﬁeld (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ .74) the behavior which we have encountered before as a gauge freedom for massless vector ﬁelds with the (free) lagrangian density L = −(1/4)Fµν F µν . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . 4 (6. As an example consider the Dirac lagrangian. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. ∂µ φ) =⇒ L (φ.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.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).73) (6. Dµ φ) − Fµν F µν .Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance.72) (6. ∂ 2 4 (6. D ∂ A (6. 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν . 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). 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. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). −e j µ Aµ = −e ρφ + e j · A. (6.

. give the equation which is satisﬁed by ψ c = Cψ .79) Exercises Exercise 6. giving the Maxwell equation coupling to the electromagnetic current. ∂µ F µν = j ν . (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. ∂µ F µν = j ν . i. −eψγ ν ψ. 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.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . where j µ = e ψγ µ ψ. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). ↔ Exercise 6.e. and deduce what is the charge of the antiparticle as compared to a particle.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds. 4 (b) What is the form of the interaction term involving a current jµ and the ﬁeld Aµ that will give the inhomogeneous Maxwell equations.1 (a) Show that the Klein-Gordon equation for the real scalar ﬁeld can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 .4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. (6.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . Exercise 6. Exercise 6.

6 (optional) Prove the properties of the tensor T µν in section 6. This relation is known as the Pauli equation. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. 2me Exercise 6. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices.4. 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. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . t) = Enr ψu (r. Comment: 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 .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 . . Hint ≡ −gs µe · B = −gs Qe s · B. t). A) (see also exercise 6. use ∂µ T µν = 0 and ∂µ M µρσ = 0.Classical lagrangian ﬁeld theory 53 Exercise 6. in order to show that Θρσ − Θσρ = ∂µ H µρσ . 2M 2M with Enr ≡ E − M ≪ M .

dq dp dq dp (7.2) H(p. dλ Examples are dO = [O. (7. there are conserved ˙ ˙ quantities Q (Eq.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q).5) ∆ x ∆3 x 54 . however. 6. q(q. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. t). (7.11) found through Noether’s theorem.3) For variations δO(p. q) = ˙ 1 mq 2 − V (q). p]P = 1. [q. Q]P . p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. labeled by the position. ˙ L(q. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. B]P = dB dA dA dB − . q) = p q − L(q. p]P .Chapter 7 Quantization of ﬁelds 7.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t).4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. (7. H]P dt and dO = [O. Further. The bilinear product for the conserved quantities. ˙ 2 (7. namely [q. starting from the lagrangian L(q. 1 qx (t) = 3 d3 x φ(x. q) also considered in the previous chapter. in our example p = mq. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. p] = i. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. is the Poisson bracket [A.

8) (7. t)] = 0. φ(y. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). qx+∆x ) . a) = Rj (Λ−1 ) φj (Λ−1 x − a).12) (7. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. (7. the discretization procedure above is an explicit example. t). py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. ∆φ.16) . • Poincar´ invariance e The full variation of the ﬁelds. 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.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . 2 2 2 (7. a) φi (x)U (Λ. t)] = [Π(x.14) In fact. Φ(f ) = d4 x φ(x) f (x).15) with furthermore the relations [φ(x. albeit with ’sharp’ functions. qx . t)] = i δ 3 (x − y). ∇φ) (7. t).7) d3 x L (φ(x).13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. ∂ qx ˙ ∂ qx ˙ (7.6) ∆3 x L x (qx . The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). i U † (Λ. discussed in the section 6. (7. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . (7. for symmetry transformations. Sometimes it is useful to keep in mind that. As an example.Quantization of ﬁelds 55 etc. Π(y. for the scalar ﬁeld theory.11) (7. φ. The essential conditions to consistently quantize a theory then are the following. ∂µ φ(x)) = ˙ d3 x L (φ. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. formally. t).9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . Π(y. ˙ δ φ(x) (7. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t).2.

(7.23) The hamiltonian in this case can be expressed in the number operator N = a† a. j (7. 2 2 (7.Quantization of ﬁelds 56 or if U (Λ. P ] 2 (7. a† ] = a† . H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ .18) with S µν given in 6. (7. 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. P ] = 0 (7. [a.25) . Q] = [P. Pµ ] = [φ (x). a† ] = 0. P ] = i. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . [Q. a] = [a† .19) 7. [N. i(xµ ∂ν − xν ∂µ )φ (x) + i (7.17) (Sµν )i φj (x). The measurements cannot inﬂuence each other or [Φ(f ). j 2 one has [φi (x). a† ] = 1. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). ω a† a + 2 2 ω i − [Q.4. φ(y)] = 0 if (x − y)2 < 0.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.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. and [N. Microscopic causality implies local commutativity. a] = −a. (7. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . sometimes referred to as second quantization. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. 2 i ωµν (S µν )i .24) It is straightforward to ﬁnd the commutation relations between N and a and a† . Mµν ] = i i∂µ φi (x).20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. Φ(g)] = 0. [φ(x).

28) (7.29) 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 .32) (7. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.26) a(k) e−i k·x + a† (k) ei k·x . which represents the ’number of particles’ with momentum k. the real scalar ﬁeld φ(x) becomes d3 k (7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. Π(x )]x0 =x′0 = 0. 7.30) [φ(x). a† (k ′ )] = 0. is equivalent with [a(k). φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. a(k ′ )] = [a† (k).e. φ(x )]x0 =x′0 = [Π(x). e. † √ a and a act as raising and lowering operators. 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).Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. i. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . (n − 1) a|n † i. but one needs to realize that in that case k 0 = Ek = k2 + M 2 . From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . Note that we will often write a(k) or a† (k).4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 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.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions.g. (7. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.like’ commutation relations between a(k) and a† (k ′ ). and (7.e.33) . (2π)3 2Ek (7. and we see that a state |0 must exist for which N |0 = a|0 = 0.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2.

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). assuring that the vacuum (by deﬁnition) has eigenvalue 0!.39) The problem with the vacuum or zero-point energy. (2π)3 2Ek (7.41) (7. (2π)3 2Ek (7. (2π)3 2Ek (7. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k).44) p|φ− (x)|0 = ei p·x . (2π)3 2Ek d3 k a† (k) ei k·x .46) (7. k 0 = Ek = k2 + M 2 ). |k = a† (k)|0 . . which now contains an inﬁnite number of oscillators.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. deﬁning particle states |k = |k (with positive energy. . This term will be adressed below. . i.35) where the vacuum contribution disappears because of rotational symmetry. d3 k a(k) e−i k·x .34) where V = (2π)3 δ 3 (0) is the space-volume. d3 k |k k|. (2π)3 2Ek (7.37) (7. The rest of the states are then obtained from the groundstate via the creation operator.40) This procedure is known as normal ordering.45) (7. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). n1 ! n2 ! (7. . in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . writing all annihilation operators to the right of the creation operators. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y).e.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).42) (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. is solved by subtracting it as an (inﬁnite) constant.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. |0 . 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. and multiparticle states |(k1 )n1 (k2 )n2 . a(k)|0 = 0. . (2π)3 (7.

d3 k = (2π)3 2Ek = (7. (7.47) ′ d3 k ′ e−i k·x+i k ·y [a(k).54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .56) = −δ 3 (x). φ− (y)] ≡ i∆+ (x − y). (2π)3 2Ek ≡ i∆− (x − y). 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 ). The last item to be checked for the scalar ﬁeld are the causality condition.18). d3 k ei k·(x−y) = −i∆+ (y − x). φ(y)] consider µν [φ+ (x). i. a† (k ′ )] 3 2E ′ (2π) k [φ− (x).e.53) t=0 . x) = 0 and hence ∆(x) = 0 for x2 < 0. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. (iii) ∆(0.48) (7.51) (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. (2π)4 (7.17) and (7. ∆(x) = − ǫ(x0 ) δ(x2 ). = − (2π)3 2Ek (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .50) or as integrals over d4 k.52) The result for the invariant commutator function is [φ(x). that they satisfy the required commutation relations in Eqs (7.49) (7. (7. φ+ (y)] d3 k e−i k·(x−y) . (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7. 2π (7. In order to calculate [φ(x). i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .

(2π)3 2Ek (7.63) i. (2π)3 2Ek (7.66) .57) (7. Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) .Quantization of ﬁelds 60 7. = ∂{µ φ∗ ∂ν} φ − L gµν . which is equivalent to the relations (only nonzero ones) [a(k).62) (7.65) 7. ↔ (7.e.61) (7. φ† (y)] = i∆(x − y). we will consider it as a repetition of the quantization procedure.60) = i φ∗ ∂µ φ. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). (2π)3 2Ek (7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (7. 2 From the lagrangian density (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) .64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x).5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). extending it with the charge operator and the introduction of particle and antiparticle operators. a† (k ′ )] = [b(k). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).59) (7. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld.

ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.74) (7.73) (7. s)ei k·x + 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. b† (k ′ . s)¯(k. s). s)e−i k·x + d† (k.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . The solution is the introduction of anticommutation relations.69) (7. (2π)3 2Ek (7. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.e. s). u v (2π)3 2Ek (7. (7.67) i ψ / ψ − M ψψ gµν .78) (7. ∂ 2 ↔ (7. s)e−i k·x . s) ei k·(x−y) x0 =y 0 .75) (7. (2π)3 2Ek d3 k b† (k. jµ Θµν = = ψγµ ψ. interchanging creation and annihilation operators. ↔ i ψγµ ∂ν ψ − 2 (7. d† (k ′ .68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). (7. s)¯(k. . then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . The quantized ﬁelds are written ψ(x) = s d3 k b(k. s)u† (k. s)v(k. † {ψi (x).Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. s′ )} = {d(k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . (2π)3 2Ek (7. i.71) (7. s) e−i k·(x−y) j s † + vi (k. s)vj (k. s)ei k·x .77) Note that achieving normal ordering. s)u(k.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. {b(k.70) where the last line is obtained by using the Dirac equation.

it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i.82) where i∆1 = i∆+ −i∆− .6 The electromagnetic ﬁeld 1 L = − Fµν F µν . −λ(∂ρ Aρ ).87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. ˙ δ A0 δL = F i0 = E i . 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. as the vanishing of Π0 induces a constraint. For the scalar combination {ψi (x). 4 2 This gives the equations of motion discussed before. ψ j (y)] = (i/x + M )ij i∆1 (x − y). ∂ (7.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . ψ j (y)} at arbitrary times one has {ψi (x).86) (7. we would have obtained [ψi (x). When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. ˙ δ Ai (7. ∂ (7.80) 3 = δ 3 (x − y) δij . 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). one has † {ψi (x). (7.83) From the lagrangian density the canonical momenta are = = (7. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.85) which reﬂects the gauge freedom. but has the problem of being noncovariant. (7.84) (7. 7.88) . 4 Π0 Πi δL = 0. which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem).

95) Exercises Exercise 7. (b) Proof for a quantity O(p.Quantization of ﬁelds 63 for physical states |A and |B . j = 1. λ)e−i k·x + c† (k.93) which does exhibit problems with the time-like photon.e. q) that dO/dλ = [O.11. pj ]P . a longitudinal photon and two transverse photons. which reproduce the Hamilton equations. i.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. 0)c(k. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). what are [ri . 2. λ)c(k. containing a time-like photon. These problems are solved by the Lorentz constraint between physical states given above. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). Aν (y)] = [Πµ (x). (7. 7. (d) Use Poisson brackets to calculate q and p. 0) . Πν (y)]x0 =y0 = i gµν δ 3 (x − y). λ). 0. and are equivalent with [c(k. where Q is the conserved quantity in Eq. λ)|A = 0.89) with four independent vectors ǫµ . (7. 6.e. The canonical equal time commutation relations are [Aµ (x). (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. λ)ei k·x . pj ]P .91) (7. Πν (y)] = 0. i. µ λ=0 (7. In fact the commutation relations imply ˙ [Aµ (x). c† (k ′ . 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. and [ℓi .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. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. for which it is suﬃcient that ∂µ Aµ |A = 0. ′ (7. 0.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). rj ]P .3) satisfy the conditions of a Lie algebra. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. [ℓi . ˙ ˙ . λ) − c† (k. (7. Q]P . It gives 3 k µ ǫ(λ) (k)c(k. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. µ λ=0 (7.94) Choosing k µ = (|k 3 |. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k.

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

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

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

9) (8. op P (8. −λ) e−ik·x − d† (k.e. λ) e−ik·˜ + d† (k. .Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. (i˜ µ ∂µ − M ) ψ(˜) = 0. −λ) e−ik·x − d† (k.5) (8. ˜ Pop d(k. x (8.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). λ)u(k. x The latter behavior of the vector ﬁeld will be discussed further below. m). x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. λ)u(k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). λ)γ0 v(k. λ)v(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.10) i. λ) P −1 = −η ∗ d(k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k.7) = λ d k x x ηP b(k. −λ).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. λ) e−ik·x − d† (k. m) γ0 v(k. In the same way as the Fermion ﬁeld. −λ)u(k. m) = = ˜ u(k.3) (8. λ)v(k. λ)γ0 u(k. iγ µ† ∂µ − M ψ(˜) = 0. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. λ) Pop = ηP b(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k.6) (check this for the standard representation. if helicity λ is used instead of the z-component of the spin m. γ x (8.11) (8. one can also consider the scalar ﬁeld and vector ﬁelds.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). ˜ −v(k. m). −λ). the above operation reverses the sign of λ). λ) eik·x . 8. −λ)v(k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k.4) (8. (2π)3 2Ek (8.

16) d k ηC d(k. m) ¯ C v T (k. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. x (8. λ).19) Cop d(k. antilinear operator is anti-unitary if A† = A−1 . The same relations hold for helicity states. i. As time reversal will transform ’bra’ into ’ket’.3 Time reversal xµ = (t. m).14) (8. 8. (8. x x (8. (8. λ) Cop = ηC d(k. λ)C uT (k. λ)C v T (k. as discussed in section 4). The (time-reversed) Dirac equation becomes. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . we start with the complex conjugated x Dirac equation for ψ. r). . it is antilinear1 . λ) eik·x . writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0.21) where A is a 4 × 4 matrix acting in the spinor space.17) 3 = λ This shows that −1 Cop b(k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. m). ((iγ µ )∗ ∂µ − M ) ψ(x) = 0.18) (8. λ) e−ik·x + b† (k.e. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . λ) −1 Cop = ∗ ηC b(k. m) ¯ = = v(k.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. λ) e−ik·x + b† (k. λ). Top |φ = φt | = (|φt )∗ . ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. r) −→ −˜µ ≡ −xµ = (−t.15) (where one must be aware of the choice of spinors made in the expansion.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . u(k. x iγ µT ∂µ − M ψ(−˜) = 0. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. Norm conservation requires Top to be anti-unitary2 . x (8.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). λ)u(k. (2π)3 2Ek (8. λ)v(k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8.

26) d k x x ηT b(k. ˜ = 0|b(k)|A 8. x (8. λ) eik·x + d† (k. λ |¯.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k.33) (8. p. λ)u∗ (k.34) . λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ)u∗ (k. λ) T −1 = η ∗ d(k. −λ |¯. The result is = = λ (8. λ) eik·x + d† (k.32) (8. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ a |a. there are 16 independent of these bilinear combinations. −λ a T a. λ) i γ5 C v(k. ∗ ˜ v (k. λ| a. λ)v ∗ (k.24) (8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. a state |a. λ| ¯ C |¯. λ).23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k.28) (8. p.Discrete symmetries 69 Table 8. op T (8. (8. −p. p. λ) eik·x + d† (k. λ)iγ5 Cv(k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k.31) (8. −p. p. λ)iγ5 Cu(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 .30) (8. 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(−˜).2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ a P |a. Since there are 16 independent 4 × 4 matrices.25) (check this for the standard representation). λ) e−ik·x (2π)3 2Ek (8. λ). −p. −p.29) In table 2 the behavior of particle states under the various transformations has been summarized. λ)v ∗ (k. λ)u∗ (k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ) eik·˜ + d† (k. λ). λ) Top = ηT b(k. λ). ˜ Top d(k. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. λ)v ∗ (k.

of which the expectation value between momentum eigenstates can be simply found. 2M (8. s . Note that the lagrangian density L (x) → L (−x) under Θ.g. ¯ In general the expectation value between momentum states can be more complicated. As an example consider the vector current for a point fermion. in this case only q 2 = (p′ − p)2 . For instance for nucleons one has Γµ (p′ . V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . 8. as well as under the combined operation Θ = PCT (see Table 3). σ µν . s′ |V µ (x)|p. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x).5 Form factors Currents play an important role in ﬁeld theory. 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 ). ¯ V µ (x) =: ψ(x)γ µ ψ(x) :... p′ . p′ |V µ (0)|p = u(p′ )γ µ u(p). γ5 γ µ or γ5 ). This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .35) The x-dependence can be accounted for straightforwardly using translation invariance.38) (8. The 16 combinations of Dirac matrices appearing above are linearly independent. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).36) (8. (8. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).37) (8. (q 2 ). called form factors. p)u(p) e−i (p−p )·x . and T.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. p′ |V µ (x)|p = u(p′ )Γµ (p′ . In many applications the expectation values of currents are needed. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. p) can be built from any combination of Dirac matrices (1.Discrete symmetries 70 Table 8. e. for the vector current V µ (x). C.40) where the coeﬃcients are functions of invariants constructed out of the momenta. γ ν ]. ¯ ′ ′ (8. γ µ . C.41) 2M 2M . 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. (q 2 ).39) where Γµ (p′ . F. F.

¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ .g. the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). e. p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p.Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. e.43) = ei q·x u(p′ )γ0 Γµ (˜′ . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . C and T invariance. p) that Γµ (p′ . µ p ˜ (8. p)(i γ5 C). p′ )u(p′ ) = u† (p′ )Γµ† (p. p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p.g. p). p) = (i γ5 C)Γ∗ (˜′ . γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . ¯ 2M (8.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . p)γ0 u(p). • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ .45) Exercises Exercise 8. ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ .1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . p)(iγ5 C)u(p). p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. p) that Γµ (p′ . or relations based on hermiticity of the operator or P . • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p′ )γ0 = Γµ (p′ . p′ )γ0 u(p). Therefore hermiticity implies for the structure of Γµ (p′ . p) = γ0 Γµ (˜′ . p) that γ0 Γµ† (p. p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p)γ0 . p)u(˜) ˜ x p ¯ p p ˜ p (8.42) where q = p′ − p. or relations that follow from the equations of motion.

0. Exercise 8.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. −|q|/2). 4M 2 = F1 + F2 . p = (E. (c) Show that hermiticity of the current requires that F1 en F2 are real. Exercise 8. 0. (b) Show that in the interaction with the electromagnetic ﬁeld. p′ = (E. 0. 0.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). 0. 0. σµν q ν F3 (q 2 ) 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 . L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M .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 . |q|). GE GM = F1 − Q2 F2 . (c) Show that if a term of this form would exist. (b) Show that such a term is also not allowed by time-reversal symmetry.

AS (t) = AS and the states o are time dependent. [AH . QH (t)|q.3) (9. t0 ) ∂t (9. ≡ 0.5) (9. t0 ) = H U (t.2) (ii) Heisenberg picture. t′ ≡ q ′ |q ′ . i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. (9. |ψH (t) = |ψH . t ′ QH (t′ )|q ′ . Consider the two (time-independent) Heisenberg states: |q.7) 73 .Chapter 9 Path integrals and quantum mechanics 9. |ψS (t) = U (t. t ′ |q . t0 ) AH (t0 ) U (t. and the operators are time-dependent. in which the operators are time-independent. (9. AH (t) = U −1 (t. t ≡ q|q. H]. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . (9.6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing.1 Time evolution as path integral U (t. t′ . t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. (9.4) ∂ U (t.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t0 )|ψS (t0 ) . t . in which the states are time-independent. t0 ).

The hamiltonian is an operator H = H(P. Then qj+1 . (9. t′ |q. 2π (9. . B]. t′ |q. Combining the two terms gives qj+1 . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . ˙ 2π 1 1 1 [A.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . [A. n |q (9. ′ (9. The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. 2 12 12 (9. tj+1 |qj .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. 2π (9. QS |q ′ ≡ q ′ |q ′ . qj ). H(P. use the Campbell-Baker-Hausdorﬀ formula. 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 | . .8) Choose t as the starting point with |q = |q.16) this. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . eA eB = eC with C = A+B+ (9..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|. Q) expressed in terms of the operators P and Q. B]] + [[A. B] + [A. Q) = K(P ) + V (Q) = (P 2 /2M ) + V (Q) one can split e−i H∆τ ≈ e−i K∆τ e−i V ∆τ .12) where the transformation between coordinate and momentum space involves q|p = ei p. B] + . . t = q ′ |e−i H(t −t) |q . qj )]) . |q1 q1 |e−i H∆τ |q .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 . tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj . tj+1 |qj .11) (9. with the correction1 being of order (∆τ )2 .14) .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 ′ . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj .qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . (9.q .

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

In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. 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. (9.21) again a multidimensional integral. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. (9.23) Having deﬁned the Gaussian integral. α(x) = n αn fn and K(x.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m. y) = ∗ m. Consider the gaussian functional as an example.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. Dα F [α] = Dα F [α + β]. 2 (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K. with Dα F [α] = n dαn N F (αn ) (9.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.25) (see Exercises). .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) (9.24) (9. Note that procedure (i) is an example of the more general procedure under (ii). N (9. A useful property of functional integration is the translation invariance.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .27) (9.

θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1.32) (9.g. Examples of functional diﬀerentiation are (9. i.35) For applications to fermion ﬁelds. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 .33) (9. δα(x) δα(x) δ αKβ = dy K(x. θη = −ηθ. θ2 = 0. αn = δmn .37) We note that from the ﬁrst to the second line one needs to realize that a pair of diﬀerent Grassmann variables behaves as ordinary complex variables. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . products of the same pair appear. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. In the expansion of the exponential (from second to third line). θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. (9.30) (9. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). α(y) = δ(x − y). F [θ. however. αy = δxy .31) δ δ α = dy α(y) = dy δ(x − y) = 1. y) θ(y) . n θn fn (x) dx θ∗ (x) θ(x) .36) where the coeﬃ- This gives for the Gaussian functional integral after expanding θ(x) = cients θn are Grassmann variables the result Dθ∗ Dθ exp (−θ∗ θ) = n ∗ dθn dθn ∗ dθn n exp − ∗ θn θn n = = n ∗ dθn exp (−θn θn ) ∗ ∗ dθn dθn (1 − θn θn ) = 1.e. dx dy θ∗ (x) K(x. (9. which vanish.29) (9. δα(x) or in discretized form δ 1 ∂ ∂ with . (9. δα(x) (9. the Gaussian-type functionals.38) .34) (9. e. In principle the deﬁnitions of functionals is the same.

Path integrals and quantum mechanics

78

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

∗ ∗ δ e−θ η = θ∗ (x) = θ∗ (x) e−θ η , δη(x) ∗ ∗ δ e−η θ = −θ(x) = −θ(x) e−η θ . δη ∗ (x)

= δ(x − y),

(9.39)

(9.40) (9.41)

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

9.3

Time ordered products of operators and path integrals

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

1 ˙ ˙ ˙ can be rewritten after rewriting the integrand as − 2 (p − q)2 + 1 q 2 − V (q) = − 1 (p − q)2 + L(q, q). 2 ˙ 2 The result is ′ t

q ′ , t′ |q, t =

Dq exp i

t

dτ L(q, q) , ˙

(9.44)

**which was considered the starting point for path integral quantization by Feynman.
**

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

t′

Dq exp i

t i 2

dτ Leﬀ (q, q) , ˙

(9.45)

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

i

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

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

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

t′ t′ t

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i

t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

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

t t′

(9.49)

dτ L(q, q) . ˙

(9.50)

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

9.4

An application: time-dependent perturbation theory

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

′

(9.51) (9.52) (9.53)

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

t′ ∞ n=0 t′

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),

t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

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

**Path integrals and quantum mechanics
**

t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

′

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

t′

U

(n)

(t , t) = (−i)

t

′

dτ HI (τ ) U (n−1) (τ, t)

t′ τ1 τn−1

= (−i)n

t

dτ1

t t′ t′

dτ2 . . .

t t′

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

t

=

(−i)n n!

dτ1

t t

dτ2 . . .

(9.60)

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

t′ t′

+

1 2

dτ2

t τ2

dτ1 HI (τ1 ) HI (τ2 ),

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

t′ t′

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )

t′

+

1 2

t′

dτ2

t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

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

t′ t′

dτ1

t t

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

the desired result.

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

∞ t′ t′ t′

dτ1

t t′ t

dτ2 . . .

t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .

t

for g. q) → L(q. t′ q ′ . T . φn (q. for the physical states o q. one has q ′ . t Q′ .5 The generating functional for time ordered products By introducing a source-term. q) + J(t) · q. t|n = q|n e−iEn t with q|n the time-independent wave function. Considering the source to be present between times t and t′ . L(q. T = n q. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. q . ˙ J (9. i. t|Q. i. T ′ |q ′ . q)] ˙ (9. n (9. t′ |q. t is related to the Schr¨dinger state |q by |q. which in turn are embedded between an early time T and a future time T ′ . t) φ∗ (Q) e+i En T . it is possible to switch on an interaction. making the electron) and the absorption of an electron (think of a detector). t′ |q. harmonic oscillator. q) + J(τ )q] .65) J dq ′ dq Q′ . t |q. t|Q. the creation of an electron (think of a radio-tube. Consider.e. T ′ |Q. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. The Heisenberg state |q. 9.64) (9.62) p Dq D exp i 2π t dτ [pq − H(p. T with q. t = = ′ ′ V = q . t q. ˙ ˙ physically pictured as.s.Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. furthermore. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p.66) . Before and after these processes there is only the vacuum or ground-state |0 .e.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. t|n n|Q. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. T = n J t′ = = Dq exp i t dτ [L(q. a set |n of physical eigenstates. t = ei Ht |q . say. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!).63) for plane waves. T < t < t′ < T ′ . q)] .

t). dq dτ + J(τ )q(τ ) . . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q.6 Euclidean formulation The above expression (Eq. Better is the use ˜ of imaginary time t = −it. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . T = φ0 (q. . in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. The importance of Z[J] is that the time ordered product of operators can be obtained from it. J (9.70) for the generating functional is in fact ill-deﬁned. t) (9. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. . δJ(tn ) hence the name generating functional.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . (9. J=0 (9. t′ |q.67) = = φ0 (q. Since we have (neglecting multiplicative factors). Q(tn )|0 . t|Q. t 0 0out |0in J . The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. 9. = (i)n 0|T Q(t1 ) . .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 ′ ). . thus T →i∞ lim q.68) (9.71) 9. t′ ) q ′ . T ′ |Q. T → i ∞ and T ′ → −i ∞. ˜ t′ → ∞.

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

2 (9. .80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η . y) K −1 (y.Path integrals and quantum mechanics 84 Exercise 9.79) (9. z) = δ(x − z).3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . where dy K(x.78) (9.

. t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. ∂µ φ) − Jφ] (10.Chapter 10 Feynman diagrams for scattering amplitudes 10. φ(x. . x′ .e.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. xn ). . . . i.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.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 ) . . . Π) (10. . t′ |φ. (10. δJ(xn ) 85 ≡ G(n) (x1 .2) = ∞ −∞ d4 xE [LE (φ. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . x. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . ∂µ φ) + Jφ] (10.4) J=0 .

(10.9) i.5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. . J(xn ).3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . . . 2 2 → (10. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. 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. (2π)4 (k 0 + iǫ)2 − E 2 (10.10) J=0 In order to determine ∆F . . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). . xn ) = x2 xn . . xn ) J(x1 ) .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). d4 xn G(n) (x1 . with Z0 [0] = 1. . Furthermore. (10.8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y).4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x .13) . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10.e. . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . As discussed in the previous section (exercise 9. . (10. . .

17) d4 k e−i k·x . (2π)4 k 2 − M 2 (10.22) (10. (10. φ+ (0)] = (2π)3 2E are also shown in exercise (7. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x.15) e−i k·x d4 k .1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10.16) C where C is a closed contour in the complex k 0 -plane. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). (2π)4 k 2 − M 2 (10.4).20) = exp − 2 where or (see ﬁg. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). (2π)4 (k 0 − iǫ)2 − E 2 (10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . ∆F = ∆C . t) = 0 for t < 0. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space.14) They are solutions of the inhomogeneous equation. 10. µ 1 1 • Firstly.18) Thus we see various solutions. 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).21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.e. t) = 0 for t > 0. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . 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). 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 . The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. i.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . 4 k 2 − M 2 + iǫ (2π) (10. (10. . In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. the advanced Green’s function ∆A (x) = satisfying ∆A (x. ∆(x) = − d4 k e−i k·x . the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .

1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility.25) x1 x2 x3 = x4 + + . 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. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. the same result as above.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). = θ(x0 − y 0 ) 0|[φ+ (x). we have G0 (x1 . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . . or diagrammatically (4) (10. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . x2 . (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. . We ﬁnd (see also section 7. because this is the only quantity entering Z0 [J]. x3 . φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x).

i. L (φ) = L0 (φ) + LI (φ). x2 . (10. x3 .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 .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 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.Feynman diagrams for scattering amplitudes 89 10. x2 ) = i ∆F (x1 − x2 ) = G(4. 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. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . (2) (0) (10.e.32) (10. (10.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. x2 .30) and in zeroth order G(0.g ) (x1 . the generating functional can be written.g 0 ) = G0 = 1. x2 ) = G0 (x1 .34) Introducing a vertex point to which four propagators are connected.27) When interactions are present.g ) (x1 .31) G(2. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. ≡ −i g d4 z (10. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. Z[J] = = Dφ exp i exp i d4 z LI (10. x4 ) = G0 (x1 .29) This expression will be the one from which Feynman rules will be derived. (10.

δJ(xn ) . did not contain parts that could be written as products of simpler Green’s functions. (10. . . d4 xn G(n) (x1 . .e. 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]. . . . d4 xn G(n) (x1 . . .36) J J from which one obtains G(0. . .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . which also was the only Green’s function for which the diagram was connected.40) in−1 n! d4 x1 . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . . (10. J=0 (10.42) x1 G(n) (x1 . (10.g 1 ) = 1 8 z 1 8 . . (10. . G(n) (x1 . . In the free case we have seen that the exponent of Z[J] contained all essential information. J(xn ) c (10.38) The result for the 4-points Green’s function is left as an exercise. J(xn ). Z[J] = n in n! d4 x1 .41) in terms of socalled connected Green’s functions. Diagrammatically this exponent only contained the two-point function. xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . . Connected Green’s functions We have now seen the Green’s functions as the quantities appearing in the expansion of the generating functional Z[J]. . . . . . (10.37) G(2. . . xn ) J(x1 ) .g ) (x1 . This remains also true for the interacting case. .39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. . . xn ) J(x1 ) . i.

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

x2 . φout and the interpolating ﬁeld φ(x). translation invariance). then (n) (n) Gsc = Gc for n ≤ 3. in|S|p′ . (10. x1 G(4. implying the absence of tadpoles. out = δαβ ↔ SS † = 1. in = p. (2) (10.47) . in = S|α. Sβα ≡ β. in|p′ . (10. in into ﬁnal state particle states (out-states) |β. this will be translated to the action on ﬁelds. (2) The one-particle state is invariant (conservation of energy and momentum. in|S = β. . . out| = eiϕ0 0. one ﬁnds for the connected 4-point Green’s function at order g. Proof: α.43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). in = |p1 . Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . Next. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . out . ∂µ ∂ µ + M 2 φin (x) = 0. x4 ) = c 1 x3 z (10. in|S = α.44) J=0 For the interacting theory. out (suppressing except momenta all other 1 m quantum numbers). out|p′ . (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). in ⇔ β. the vacuum expectation value of the ﬁeld φ(x) is zero. p′ . .e. in = p. For the free scalar theory.45) x2 as the only surviving diagram (see exercises).3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. e. out|α. in| (choose ϕ0 = 0). in = |α. Proof: 0. in|SS † |α. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). out = |p′ .Feynman diagrams for scattering amplitudes 92 i. pn . out| ⇔ |α. considered explicitly.g ) (x1 . in = p. . out|α. 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 . thus β.g. x4 10. (10. x3 . p. . out = p|p′ . out| and S † |α. in = β. . out|p′ . in|S = 0.

(2π)3 2E (10. out|φout = β.52) where Z is a constant. (10. z) φ(z).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. (10.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. √ Although the above. β.and out-ﬁelds is: φin (x) = S φout (x)S −1 .g. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. z) φ(z) d4 y d4 z ∆A (x − y) K(y. the source term Jφ in the lagrangian density is treated in the interaction picture.51) (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. a)SU −1 (Λ. a) = S. 2 ∂µ ∂ µ + M0 φ(x) = J(x).54) . S is invariant under e Poincar´ transformations: U (Λ.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . i. in|φin S → φin S = S φout (x). φ(y)] = Z [φin (x). as it stands. a causality condition that can be proven to imply the absence of interactions. Proof: β. The time evolution operator (see Eq. which in a later stage (renormalization) will become more important. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. e. (10. The relation between S and Z[J] To establish this relation. implies the strong (operator) convergence φ(x) → Zφin (x). this can actually not be used as it would imply [φ(x).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 ). in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. 9.e. The relation between S-matrix and in. φin (y)] = iZ ∆(x − y). The convergence therefore must be weakened to t→−∞ √ (10.53) 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). φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .

(10. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. just as we did for φ(x). B + C]e e . δJ(z) (10.62) where F [J] is some (arbitrary) functional. we can. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.59) (10. C] are c-numbers. provided [A. one has [A. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. B] (for the case that [A. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. [A. δJ(z) (10.58) In order to ﬁnd a solution to this equation note that. (10. φ(x).60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). 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(y)] f (y) e e . Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). z) . B] and [A. B]eB . B] is a c-number). δU [J] δJ(x) = = √ i Z U [J] φin (x) − √ i Z φout (x) U [J] − d4 y d4 z ∆R (x − y) K(y.e.61) e φ+ f where the normal ordered expression : e φ f : is used. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. e = i.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. z) δU [J] . t i Ut∞ [J] φ(x)U−∞ [J] (10. eB ] = [A. z) d4 y d4 z ∆(x − y)K(y. φ− f (10.63) . φ(x). δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10.57) i [φin (x).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). S U [J]] = √ Z δ S U [J]. e e ] = [A. express it in terms of advanced and retarded Green’s functions.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. y) δ δJ(y) : F [J]. which is equal to the expression e in which the creation operators are placed left of annihilation operators.

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

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

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

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

1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . 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. Vector ﬁelds and Quantum Electrodynamics As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). 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 1 ∂ (10. λ (∂µ Aµ )2 2 1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . 4 2 In addition to the fermion propagator and fermion wave functions discussed in the previous section we have the photon propagator1 µ k ν = −i k2 gµν 1 − 1− + iǫ λ (k 2 kµ kν . 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. Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . + 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. The wave functions are given by µ k ǫµ (k) k µ incoming photon outgoing photon ǫ∗ (k) µ The vertex for the coupling of photon to the electron is given by µ = i j −i e(γµ )ji .Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop.

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

Exercises Exercise 10. . s4 )γ ν u(p.s4 = 1 2s × v (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 .s2 . . s1 )|2 ¯ s 1 . s1 ) u(k.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. the masses. s2 )γµ u(k. s4 ) u e2 v (k ′ . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. s2 )γµ u(k. s3 )γ µ v(p′ .s2 3 . 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. s1 )γν v(k ′ . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s2 ) ¯ ¯ 1 2s u(p.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . (c) Do the same for the connected Green function Gc . (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive.2.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . (4) (4) . s3 ) ¯ ¯ 1 . s3 )γ µ v(p′ . .s3 . s4 ) v (p′ . The diagram. Similarly. s2 (10. This is known as crossing symmetry. 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. To lowest order (∝ α = e2 /4π) only one diagram contributes.

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

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

m2 .11) (11. p2 ) = = where λ12 denotes λ(s.12) pb = (mb . m2 . Also in this case one can express the energy and momentum in the invariants.9) (11. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . m2 ) a b . 2mb (11. ptrf ). = 2 s λ(s. p1 . (11. m2 ) is a function symmetric in its three arguments. m2 ). trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b .10) The function λ(s. In that case trf pa = (Ea . a (11.15) Explicit calculation of the reduced two-body phase space element gives dR(s. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . 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. for instance.13) (11. 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = .8) (11. m2 ) a b . m2 . which in the speciﬁc case a b also can be expressed as λ(s. p2 ) = CM = = 1 (2π)2 1 (2π)2 1 (2π)2 d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2 which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s. p1 . 4π s 4π 8π s 4π (11. 2 s s − m2 + m2 a b √ . 0).16) . 4s (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 ) = 4(pa · pb )2 − 4p2 p2 . mb and s). 2mb λ(s. use the ﬁrst relation and the abovementioned threshold value for s to calculate the threshold for a speciﬁc n-particle ﬁnal state in the target rest frame. m2 .14) One can.

25) dR(s. Consider the 1-particle inclusive case.21) (11.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt.4). pd ) = 4π s 4π 8π s 4π dt dt √ √ . for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . pz = mT sinh y. Consider the process a + b → c + d. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.20) → π0 + n → π+ + π− + n → . H1 + H2 → h + X. there is usually a preferred direction. At high energies. 11. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. Instead of the scattering angle. Writing E = mT cosh y. pT = (px .Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. . pc . = (11. 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 . a c with q = λab /4s and q ′ = being 0 or 180 degrees.19) (11.. The initial state.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. A speciﬁc reaction channel starts contributing at the threshold value (Eq. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. The various possibilities are referred to as diﬀerent reaction channels.18) (11..27) with m2 = m2 − p2 = m2 + p2 + p2 . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . in which one particle is detected. usually is a 2-particle state.22) (11. however. py ). 2s (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 . (11. in exclusive measurements all particles are detected. An often used set of invariants are the Mandelstam variables. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).17) (11. usually many particles are produced. tmax min λcd /4s. E . The variable s is always larger than a c b d the minimal value (ma + mb )2 . In inclusive measurements no particles are detected in the ﬁnal state. Of course there are not only 2-particle channels. Examples appear in → π− + p (11.

This is known as crossing symmetry. 11. u = (p1 − p4 )2 = (p2 − p3 )2 . (11. ut = u) = Mab→cd (s. t.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1.30) (11. c bd Mad→c¯(su = u. t. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t.2 Crossing symmetry In the previous chapter. tt = s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). u). (11. ¯ b (11. uu = s) = Mab→cd (s. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u.28) which (only involving angles) is easier to determine.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. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. t = (p1 − p3 )2 = (p2 − p4 )2 . u) with s = (p1 + p2 )2 = (p3 + p4 )2 . t. tu = s. ¯ tt = (pa − p¯)2 = s. η = − ln(tan(θ/2)) = tanh−1 (cos θ). b ut = (pa − pd )2 = u. ¯ tu = (pa − pc )2 = t. u). 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. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. which means that rapidity distributions dN/dy maintain their shape. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. uu = (pa − p¯)2 = s. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = .31) .

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

N 1 . .33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). For the (proper) decay width one thus has 1 N Γ0 = . (11. σ= √ 2 λab T · V (11.36) (11.35) This quantity is not covariant. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ.T V.34) i. . . (11. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. (2π)4 δ 4 (Pi − Pf ) 2 = (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf ) d4 x ei(Pi −Pf )·x V. The 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 .T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). Microscopically one has Ndecay = Na · Γ T or Γ= N 1 .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. .Scattering theory 108 Although this at ﬁrst sight does not look covariant. Using ρtrf = a (0) (0) (0) lab lab lab ρa · γa = ρa · Ea /ma (where ρa is the rest frame density) and va = plab /Ea we have a trf ρa ρb va = (0) ρa ρb 2 λab 4 ma mb (0) (0) or with ρa = 2 ma . as expected. pn ).38) (in which we can calculate −iMf i using Feynman diagrams).e. . dt (11. For this we employ in essence Fermi’s Golden rule stating that when the S-matrix element is written as Sf i = δf i − (2π)4 δ 4 (Pi − Pf ) iMf i (11. N and T · V are covariant. it is. T · V ρa (11.

16π 2 s (11. respectively. p1 . (11.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . (11. . θ) or M (q i . . θ) = −8π s ℓ (in analogy with the expansion for f (E. . θ) in quantum mechanics. for which the √ amplitudes squared have been calculated in the previous chapter. ˆ ˆ Inserted in the unitarity condition for M . . .Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). 11. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). q f ). θ). i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. . δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . see 11. θcm ) √ q 8π s 2 2 (11.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix.e. .43) dΩcm = π M (s. p1 . . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. i. The result is N = |Mf i |2 dR(s.47) M (s.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. . t) λab 4π dt.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. note the sign and cos θ = qi · qf ). . .41) (11. q n ) q n qn √ Mni (q i . p1 . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . (11.40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . θ)|2 . implies for the scattering matrix M .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. pn ). q n ).42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. pn ). (11. (11. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . for which one has dσ/dΩ = |f (E. 8π s 2π 8π s . 2 λab and for instance for two particles dσ = q ′ M (s.

Using σ= dΩ q ′ M (s. θcm ) √ q 8π s 2 . in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). 2i q 2i q (11.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . n (11.49) or Im Mℓ = q |Mℓ |2 .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 .50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. (11. In general a given channel has |Sℓ (s)| ≤ 1. 0).51) and for the case that this is the only channel (purely elastic scattering. 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 .52) From the imaginary part of M (s. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . (11. (11.54) ℓ . the total cross section can be determined. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). (11. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)).e. ℓ (11. q ˆ i.53) The diﬀerence is the inelastic cross section.

the time-evolution operator.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. The prescription for the pole structure. . 0). It is important to note that the physical width of a particle is the imaginary part of the two-point vertex at s = M 2 . 0) ∝ e−i(E−iΓ/2)t . For an unstable particle one expects that U (t. The product of amplitudes is of order e4 . (11. however. (q Mℓ )ij (s) = −M Γi Γj .58) . The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 .55) (note that we have disregarded spin). .e. We note that unitarity is generally broken in a ﬁnite order calculation. For the amplitude in a scattering process going through a resonance. it is straightforward to write down the partial wave amplitude. i.56) (again disregarding spin). To check unitarity we need the amplitude at order e4 . 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 . σel = σinel . . in which case. . 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.e. 0)|2 = e−Γt . i. ∝ U (t. but still real. such that |U (t. The quantity Γ is precisely the width for unstable particles. s − M 2 + iM Γ (11. 11. pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. For instance the tree-level calculation of Feynman diagrams at order e2 was real. This is what we will do to discuss decay widths. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. which is given by a sum of the partial widths into the diﬀerent channels.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. η = 0.57) This shows that Γ is the width of the resonance. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. This is (somewhat sloppy!) seen by considering the (amputated) 1PI two-point vertex Γ(2) = −i ∆ as the amplitude −iM for scattering a particle into itself (forward!) through the decay channels as intermediate states.

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

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. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. 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. ρcm ρcm a b 1 cm .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∆ .999 877 Γ and Γ(π − → e− νe ) = 0. Exercise 11.8.2 Calculate the ﬂux factor.000 123 Γ.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Exercise 11. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Compare the results with the experimental π − lifetime τ = 2. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. 4.2). Thus.Scattering theory 113 Exercise 11. Calculate now fπ . s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . writing N (s) M1 (s) = 2 + iM Γ . cm |va − vb | in the center of mass system for two colliding particles a and b.

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

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

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

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

inf

(12.3)

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

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

The standard model

115

**which implies W ′ = U (θ) W µ U −1 (θ) − µ or inﬁnitesimal
**

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

a a b c Ga = ∂µ Wν − ∂ν Wµ + g cabc Wµ Wν , µν

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

**QCD, an example of a nonabelian gauge theory
**

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

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

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

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

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

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

**A geometric picture of gauge theories
**

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

1 The

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

λ7 =

i

−i

1 1 λ8 = √ 3

−2

∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. -dx -dy µ µ dyµ µ dx For a vector. should be independent of such transformations. 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 . We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x).15) ψ(x).The standard model the basis for x G and x+dx 117 G. 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. 12. which connects two identical vectors. (12. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). (12.16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). (12.19) This gives rise to a (path dependent) phase in each point. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). However. The relation in Eq. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. = . ds s dxµ ds′ P exp ig A (s′ ) ψ(0).e. but expresses them with respect to diﬀerent bases. i. (12.16.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. ds which is solved by dψ(s) ds = dxµ . In principle such a phase in a given point is not observable. i. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0.14) (12. ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0).

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

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. The situation now is the following. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc .) terms constitute interaction terms. therefore a choice must be made.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. Any constant ﬁeld with ’length’ F is a possible groundstate. In this case the spectrum is described . The classical groundstate appears degenerate. . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. (12.23) F +η the latter only forming a minimum for m2 < 0. . The η massless particles are called Goldstone bosons. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. It is only invariant under rotations around the 3-axis. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. They are known as the Weyl mode or the Goldstone mode: Weyl mode. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . .21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero).e. i. for the vacuum Qa |0 = 0. ϕ2 and η. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . etc.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . has less symmetry. 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. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. say 0 0 . (12. while the quadratic terms must correspond with non-negative masses. while the lagrangian including the nonzero groundstate expectation value chosen by nature. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. 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 . In the case of degeneracy.25) 2 2 2 Therefore there are 2 massless scalar particles. In this situation one speaks of spontaneous symmetry breaking. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . (12. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0.

A bit more formal. Taking the derivative. V µ = ψγ µ T ψ. a and a′ are degenerate states. This means that they are operators that create states from the vacuum.The standard model 120 by degenerate representations of the symmetry group. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). and one must have mπa = 0. it is easy to see that one gets ∂µ V µ = i ψ [M. (12. [Qa . T ] ψ. strange. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space.. (12. there must be a nonzero expectation value 0|Jµ (x)|π (k) . In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. down. π (12.e. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab .29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor).e.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. M = . . a ∂ µ Jµ = 0. ψ −→ ei α·T ψ. 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. i. H] = 0.32) (12.) one can write L = f ψ f (i/ − Mf )ψf . ψu mu md ψ = ψd . ∂ ∂ (12. . Including a sum over the diﬀerent ﬂavors (up. Using the Dirac equation. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . Neglecting at this point the local color symmetry.e. 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. Goldstone mode. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks.31) . denoted |π a (k) . π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. a massless Goldstone boson for each ’broken’ generator. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. The generators Qa (in that case the rotation operators Lz . i. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). = ψ(i/ − M )ψ. T = τ /2.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. i. called ﬂavors. 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′ . if the generators Qa generate a symmetry. .26) for all the states labeled by a corresponding to ’broken’ generators. . etc.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). (12. irrespective of the fact if they annihilate the vacuum.

35 is nonzero. or. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . From group theory (Schur’s theorem) one knows that the latter can only be true. socalled axial vector transformations. generated by TR/L = 1 (1 ± γ5 )T . T γ5 = τ γ5 /2. In the real world. i.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. . That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. where the quark masses are not completely zero.e. 12. is what happens in the real world. the lagrangian density is invariant under left (L) and right (R) rotations independently. This symmetry. T ] = 0. ∂ ∂ (12. (12. This situation. for which the ﬁeld (according to the discussion above) has the same behavior under parity. ψR/L = 1 (1 ± γ5 )ψ. a doublet (isospin 1/2) of nucleons. (leaving out the ﬂavor structure) the same as ψγ5 ψ. proton and neutron. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . This combined symmetry is what one calls chiral symmetry. There exists another set of symmetry transformations. M = m · 1. etc. (12.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. chiral symmetry is not perfect. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . etc. parity minus). How can we see this. i.e. it is easy to see that one gets ∂µ Aµ = i ψ {M.e. fπ = 0 remains. if in ﬂavor space M is proportional to the unit matrix. stated equivalently. 12. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . but the fact that the symmetry is not perfect and the right hand side of Eq. but also a triplet (isospin 1) of pions. behaves as a pseudoscalar particle (spin zero. T } γ5 ψ. e. I. Note that these transformations also work on the spinor indices. including in Dirac space a γ5 matrix.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. Aµ = ψγ µ T γ5 ψ. which is approximately true for the up and down quarks. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. as the quantity ∂µ Aµ (x). (12. is by nature not realized in the Weyl mode. i. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity.34) Using the Dirac equation. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. 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. 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. which is equivalent 2 to the V-A symmetry. both are very small.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass.27. while the groundstate is only invariant under isospin rotations (R = L).g. much smaller than any of the other mesons or baryons. however. parity would not play a role in the world.e.e. Therefore the world of up and down quarks describing pions.3 MeV/c2 and Mn = 939. with almost degenerate masses (Mp = 938. ψ −→ ei α·T γ5 ψ.6 MeV/c2 ). gives also rise to a nonzero mass for the Goldstone bosons according to Eq.

one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds.39) (12. thus 9 independent degrees of freedom. Initially there are 3 massless gauge ﬁelds (each. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . We have the possibility to perform local gauge transformations. like a photon. After symmetry breaking the same number (as expected) comes out. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle.The standard model 122 12. . 0 φ= (12.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . (12.41) 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 . but one has 1 massless gauge ﬁeld (2). in this case three-dimensional) representation reads (La )ij = −i ǫaij . made into a gauge theory. . having two independent spin components) and three scalar ﬁelds (one degree of freedom each). Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation.37) (12. . again 9 degrees of freedom. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). One may wonder about the degrees of freedom. 4 2 where Since the explicit (adjoint.40) a Dµ φ = ∂µ φ − ig Wµ La φ.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. as in this case there are no massless Goldstone bosons. Dµ φ = Gµν = ∂µ φ + g Wµ × 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 ). (12. . is made. (12. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). The symmetry is broken in the same way as before and the same choice for the vacuum. 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 + . Note that the number of massless gauge bosons (in this case one) coincides with the number of generators corresponding to the remaining symmetry (in this case rotations around the 3-axis). The diﬀerence comes when we reparametrize the ﬁeld φ. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. i. while the number of massive gauge bosons coincides with the number of ’broken’ generators. 0 ϕc = 0|φ |0 = 0 . 0 0 0 = .e.

the leptons (elektrons. 1 As they are massless. The procedure. (12. R −→ e−i β R.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. U(1)Y U(1)Y (12. its corresponding neutrino (νe . In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. that gives a good description of the physical world. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist.and righthanded particles. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.). left. muon µ− or tau τ − ). = ∂µ L − (12.44) which has an SU (2)W symmetry under transformations eiα·TW . implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment.51) . the quarks (up.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.e. etc. etc.45) (12.48) (12. (12. i.) and the gauge bosons responsible for the strong. 2 (12. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . νµ and ντ ) and their antiparticles. R.The standard model 123 12.46) −→ 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. 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. however. ψR/L = 2 (1 ± γ5 )ψ decouple. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained.47) and YW is considered as an operator that generates a U (1)Y symmetry. 2 2 = ∂µ R + i g ′ Bµ R. electromagnetic and weak forces. This is written as 3 Q = TW + YW . explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. ∂ ∂ ∂ (12. while the lefthanded particles form a doublet. explicitly L −→ e−i β/2 L. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− .50) (12. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. neutrinos. down. muons. τ ).

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 = 0.56) 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. h) (12.59) . . (12. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . Putting this in leads to L (f ) =L (f 1) +L (f 2) .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. For the classical ﬁeld the choice θ4 = v is made.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. (12. −V (φ) L and (h2) = −Ge (LφR + Rφ L). .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. † i i g Wµ · τ − g ′ Bµ )φ. 4 4 In order to break the symmetry to the symmetry of the physical world. leading to the parametrization 1 0 φ(x) = √ (12. L where L (h1) (h) =L (h1) +L (h2) .58) (12. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . Using local gauge invariance θi for i = 1. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). this leads to the lagrangian L (h1) . + 8 (12. (12. . the U (1)Q symmetry (generated by the charge operator). with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. . .52) L L (f 1) (f 2) = = i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ + i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 .

69) .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 .63) (12. 2 Zµ (12. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.67) The coupling of electrons or muons to their respective neutrinos. (12.60) (12.65) The weak mixing angle is related to the ratio of coupling constants. and correspond to three massive particle ﬁelds (W ± and Z 0 ) and one massless ﬁeld (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 . 4 2 MW g2 v2 = .66) From the coupling to the photon. 2θ 2θ 4 cos W cos W 0. (12. 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 .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 . 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 . g ′ /g = tan θW .64) (12. for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + . . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2 (12. .61) (12.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .

80) First. 1.75) 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 . g2 MZ 2 2 (CV + CA ).73) (12. Comparing this with the total width into (invisible!) channels. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).72) (12.231 2.71) GeV −2 = = = = . 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. . muon or tau. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . me /v is extremely small. 91.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. 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.5 MeV (exp. (12. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. v (12.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. Finally we want to say something about the weak properties of the quarks.g. −5 (12. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).e.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 2 2 2 (12. from the MW mass is about 250 GeV. (12. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).74) (12. giving rise to a running coupling constant after renormalization of the ﬁeld theory.19 GeV. The coupling to the Higgs particle is weak as the value for v calculated e.40 GeV.e. (12. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. g ′ and v determine a number of experimentally measurable quantities. one has the relation g2 e2 1 GF √ = . 80. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. (12. i. i.166 4 × 10 0.77) (12. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.76) with CV CA 3 = TW − 2 sin2 θW Q. With the choosen vacuum expectation value for the Higgs ﬁeld.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. 3 = TW . For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.

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

. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks.34. (12. magnitudes of the entries in the CKM matrix are nicely represented parametrization 1 − 1 λ2 λ λ3 A(ρ − i η) 2 1 2 −λ 1− 2λ λ2 A V = 3 2 λ A(1 − ρ − i η) −λ A 1 12. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . the weak mass eigenstates are L (h2. e.82 and ρ ≈ 0. UR = uR cR tR .g. † Λu Λ† = Vu G2 Vu . Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. d d (12.82) space. u u where we include now the three lefthanded quark doublets in QL . allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .86) the unitary CKM-matrix introduced above in an ad hoc way. respectively).227.e. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . i.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. The Higgs ﬁeld is still limited to one complex doublet. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .22 and η ≈ 0.5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0.83) where Vu and Vd are unitary matrices. (12.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. 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 . each of these containing the three families.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . A ≈ 0. ms and mb ). 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 .

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

the left. 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. are equivalent to two decoupled Majorana neutrinos (see below).The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section.94) M L NL NL + M L NL NL . couple the right.c.95) 2 In order to have more than a completely decoupled sector.99) . (12. Absorption of phases in the states is not possible for Majorana neutrinos. α2 and δ).ν =− containing three (CP-violating) phases (α1 . The mass matrix can be written as ML |MD | eiφ M = (12. a unitary matrix relating them to the weak eigenstates. 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. For these light Majorana neutrinos one has. (12. Thus (disregarding family structure) one has two Majorana neutrinos. We use here the primes starting with the weak eigenstates. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass.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. 1 L Υ ′ = nR + nc . The way to circumvent this is to introduce as in the previous section righthanded neutrinos. one must for the neutrinos as well as charged leptons. is to add in Eq. 12. coupled by a Dirac mass term.and righthanded species then just form a Dirac fermion.100) |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2 1 nc nR M L M D nL L c + h. 1 ∗ c c (12.ν = − MR NR NR + MR NR NR . For the leptons. as above. 1 c ∗ c L mass. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. hence the mixing matrix becomes iα /2 0 0 e 1 (12. however. one being massive.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. This is called the see-saw mechanism (outlined below). with for the righthanded sector a mass term MR . − MD MR nR 2 (12. Thus 2 Υ ′ = nc + nL . For the neutrino sector. 2 R ψ = nR + nL . 0 0 1 L mass. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. the massless and massive Majorana neutrinos.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . Actually. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).98) (12.

. T } γ5 ψ. (12. related via c Υ1R = U nL . Prove that if the generators Qa generate a symmetry. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. (12.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 .103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . 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).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 .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)) . Exercise 12. H] = 0. i ψ {M. i. Exercises Exercise 12. 2 with real masses Mi .1 Consider the case of the Weyl mode for symmetries. i. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR .The standard model 131 taking ML and MR real and non-negative. a and a′ are degenerate states. Υ2R nR (12.e. Exercise 12. L Υ2L nR Υ1 and Υ2 .2 Derive for the vector and axial vector currents. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . [Qa .105) for i = 1. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. 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′ .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12.e. T ] ψ. 2 U (M M † ) U † = M0 . (12.

Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 .The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . The mass of the Z 0 is MZ = 91 GeV. Mf M2 and ghf f = . Exercise 12. ZZh. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . . Exercise 12. hhh. Exercise 12. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . the weak mixing angle is given by ¯ sin2 θW = 0. etc. e+ e− h. (b) Calculate the width to electron-positron pair.231.7 In this exercise two limits are investigated for the two-Majorana case. and the width to a pair of neutrino’s. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . 3 TW . to be precise one has for bosons (b) or fermions (f). mb ≈ 5 GeV and MW ≈ 80 GeV.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian.) are proportional to the mass squared for bosons or mass for fermions of the particles.5 Show that the couplings to the Higgs (for W + W − h. Γ(Z 0 → e+ e− ). Γ(Z 0 → νe νe ). 2 2 The masses are mt ≈ 175 GeV. ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld. at least up to a high (which?) order in λ. Exercise 12.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix.

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