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

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

Other text books of Quantum Field Theory that are useful are given in refs [3-6]. Corresponding chapters in books of Ryder.4 3.2 3. 3. Quantum Field Theory.1 4. S. Quantum Field Theory.2 1.3 3. M. The notes contain all essential information.3 References As most directly related books to these notes. Weinberg.3 2. Srednicki.V. The quantum theory of ﬁelds.5 2.2. 3.5 5.3 6. Vol. Itzykson and J.3 3. I: Foundations.2 36 22 . Vol. 4. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012). 6. Peskin & Schroeder and Srednicki.1 3.2 3. 2007.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. 1985. L.2 2.2 3. 1995.2 2.2 3. Cambridge University Press. An introduction to Quantum Field Theory. Diagrammatica.4 3.8 3.4 3. 5.5 4.2 3.1 2 2 3.3 2. M. 1995. Cambridge University Press. Cambridge University Press.1.1 6.H.4 6. I refer to the book of Srednicki [1] and Ryder [2].2 2.3 4. Quantum Field Theory. but are rather compact.3 2. 1996. Veltman.2 6. Cambridge University Press.3. Zuber.3 2.E. Peskin and D. McGraw-Hill.4 2. 2.2 6.7 2.3 3. Addison-Wesly. Schroeder. Cambridge University Press. 2.4 4.-B. These notes Section 1.1 5.2 4. M. C.5 2. 1980. 1.4 2. Ryder.3 3.3 3.1 3.1 2. 1994.1 2. 3.1 1.3 2. II: Modern Applications.

4 9.4 12.1 12.5 9.1.4 3.2.1 20. 4.5 4. 6.5 9.2 10.1 7.4 2.1 20.4.1 20.6 3. 2.1 9.3 12.2 8.5. 6.4 4.6 3. 20.5 4. 2.5 6.1.4 These notes Section 7.4.3 7. 4.4 10.3 20.7 4.8 6.6 8.7.2. 6.6 4.3 7.1 8.3 11.1 11.7 5.3 6.2 9.3.2 8. 9.5 2.3.8 5 4.3 9.1 5. 4.10 8.6 9.2.6 6.2.3 8. 4.4 12.5 7.2 12.6 10. 8.2 5.1 9.6 3.2 7.1 10. 6. 9.2 4.4.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.3 6 8 9 Srednicki 3 3 3 37 5.3 10.5 6 7 6 6.5 .3.3 3. 6.4 7.3 8.5 11.2 11.5 7.4 8.3 2.1.

• Slight rewriting of discussion of Majorana mass term in section 6.3 • First two sentences of Chapter 5 have been modiﬁed.5 Corrections 2011/2012) • Extension of Exercise 3.2 .

6) .1) In the limit that the action S is much larger than . the action of the momentum operator in the coordinate representation is not as simple as the position operator. (1. the position can in principle be determined at any time with any accuracy. usually given divided by 2π.054 571 68 (18) × 10−34 J s = 6. quantum eﬀects do not play a role anymore and one is in the classical domain.Chapter 1 Introduction 1.5) One can also choose a representation in which the momenta of the particles are the dynamical variables.2) In the limit that v ≪ c one reaches the non-relativistic domain. ˜ ψ(p) = i d3 r exp − p · r ψ(r). In quantum mechanics. Indeed. (1. being eigenvalues of the position operators. It is given by pop ψ(r) = −i ∇ψ(r).4) where δij is the Kronecker δ function. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. In special relativity a special role is played by the velocity of light c. One can talk about states |r and the wave function ψ(r) = r||ψ . They satisfy the well-known (canonical) operator commutation relations [ri . (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. ≡ h/2π = 1. 1 (1. c = 299 792 458 m s−1 .1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). Corresponding position and momentum operators do not commute. (1. In quantum mechanics a special role is played by Planck’s constant h. (1.582 119 15 (56) × 10−22 MeV s. pj ] = i δij . S ≫ .

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.

x2 .24) The quantity gµν ≡ nµ · nν is the metric tensor. xi are the three space components. in which case zi = ǫijk xj yk . one must distinguish tensors with upper or lower indices and one can have mixed tensors.3). x3 ) = (t.2. One has x2 = xµ xµ = t2 − x2 . When 3-dimensional space is considered as part of Minkowski space. Unlike in Euclidean space. x3 ). −x). So we could have used all upper indices in the above equations. x1 . we will use upper indices for the three-vectors. In special relativity we start with a four-dimensional real vector space E(1. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations. ˆ ˆ (1.1. it is convenient to distinguish lower indices from upper indices.23) δjm δkn − δjn δkm . x2 . No diﬀerence is made between upper or lower. We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. x) = (x0 . linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations.25) Within Minkowski space the real. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. The length (squared) of a vector is obtained from the scalar ˆ product. • ds2 < 0 (spacelike intervals). (1. The real. Because of the diﬀerent signs occurring in gµν . (1.νn can be conν structed.. δim δjm δkm δin δjn δkn . One can distinguish: • ds2 > 0 (timelike intervals). Relations relating tensor expressions. the invariant length or distance (squared) is not positive deﬁnite. Many other four vectors and tensors transforming like T ′ µ1 . Λµn n T ν1 .3) with basis nµ (µ = ˆ 0.26) (1. 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 lower indices are constructed in the following way. The (invariant) lengths often have special names. denoted as V ′µ = Λµν V ν .21) (1. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . Vectors are denoted x = xµ nµ . • ds2 = 0 (lightlike or null intervals). xµ = gµν xν . x2 = x · x = xµ xν nµ · nν = xµ xν gµν . in this case an inertial system exists in which the two points are at the same time and ds2 just represents minus the spatial distance squred ds2 = −dx2 . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). The Poincar´ transformations include e e Lorentz transformations and translations. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. where the coordinate t = x0 is referred to as the time component. . however. the points lie on the lightcone and they can be connected by a light signal. independent of a coordinate system. are ... . and are given by (x0 . x1 .22) (1.µn = Λµ1ν1 . These transformations do change the components of a vector.Introduction 5 that can be used in the cross product of two vectors z = x × y. 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 . In Minkowski space.

Introduction 6 called covariant.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used).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. ∂2 .34) (1. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. 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. for instance.36) Exercises Exercise 1. (1. but more important has the same eﬀect on lefthandside and righthandside. because one has (1. p) = (E. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . (1. a∞ = 4πǫ0 2 /me e2 . where m is called the mass of the system.28) ν ν Note that gµ with one upper and lower index. Each of these permutations leads to a minus sign. It is an invariant tensor.27) It is easy to convince oneself of the nature of the indices in the above equation. (1. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. Note that in this equation one has on left. (1. = −24. ∂t (1. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . . For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .32) .33) (1. ∂3 ) = ∂ ∂ ∂ ∂ .∇ . ′ ′ .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. . 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. ∂1 . g0 = g1 = g2 = g3 = 1. explicitly written as (p0 . p). starting from ǫ0123 = 1. (1.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . The length squared of ∂ is the d’Alembertian operator. constructed via the metric tensor itself. not aﬀected by Lorentz transformations.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. deﬁned by ∂2 − ∇2 . 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 . = ∂ .

511 MeV. 1. n+ . g−− . g+− and g−+ . ]. One does not need to know . Also for the coordinates [a− . 0) ˆ n3 = (0. Also construct and calculate the Planck length Lpl . n3 . n1 . ˆ ˆ ˆ ˆ These are n1 = (0. a3 ) are the expansion coeﬃcients using the basis vectors n0 .3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ .71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). This demonstrates how a careful use of units can save a lot of work. by a simple few-line reasoning [For instance: If {µ. ν. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. 1. a+ . but only appropriate combinations. ǫ0 . which is the Compton wavelength for the Planck mass.(A · B) C using the properties of the tensor ǫijk given in section 1. . 1) ˆ . Compare it with the proton mass and use Eq. a2 ] we can ﬁnd basis vectors n− . 3} the index α can only be equal to one of the indices in ǫµνρσ . n2 . Exercise 1. a1 . . a2 ] or [a− . ˆ ˆ . a+ . 0. α and me c2 = 0. 2. [Note: what is the MKS unit for V/T?] Exercise 1. 0. 0) ˆ n0 = (1. Exercise 1. n1 . 0. a2 . 1. How are a+ and a− related to a+ and a− . 0.2 Prove the identity A × (B × C) = (A · C) B . (b) The coordinates (a0 . 0. e. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . n2 . σ} is a permutation of {0. 0) ˆ n2 = (0. c. (d) Use the value of the gravitational constant GN / c5 = 6. 1. re = e2 /4πǫ0 me c2 to the Compton wavelength. . 0.13 to give its actual value in kg.3. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . ρ.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . (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). (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. a+ . me . a1 .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). a1 .

p).4) where M is the particle mass.1 The Klein-Gordon equation In this chapter. t) = − ψ(r.2. namely plane waves characterized by a wave number k. t) = exp(−i k 0 t + i k · r). we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. E = p2 . the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite. ∂t 2M (2. ∂t2 (2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . i ∂ ∇2 ψ(r. (2. (2.2) Acting on the wave function one ﬁnds for a free particle. ∂t p −→ pop = −i∇. φk (r. But the replacement 2. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. 2M (2.3 are not covariant. (2. p2 = pµ pµ = E 2 − p2 = M 2 . t). one obtains (2. 2 + M 2 φ(r.Chapter 2 Relativistic wave equations 2.5) with (k 0 )2 = k2 + M 2 . 8 . The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.7) k 0 = ± k2 + M 2 = ±Ek . Although it is straightforward to ﬁnd the solutions of this equation. t) = 0.3) Equations 2.1 and 2. t) = ∂2 − ∇2 + M 2 φ(r. written as pµ −→ i∂µ is covariant (the same in every frame of reference).

And. k) + ei k·x φ(−Ek . (2π)3 2Ek (2.9) They satisfy the continuity equation. there are solutions with negative energy. relativistically the density is not a scalar quantity. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. (2.8) (2. but rather the zero component of a four vector. j) of the four-current j.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. 2.11) Therefore. ∂ρ = −∇ · j. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . As we will see later both problems are related and have to do with the existence of particles and antiparticles. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. ↔ ↔ (2. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. This operator has all possible solutions in it multiplied with creation (and annihilation) operators.e. for which we need the interpretation of φ itself as an operator.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds.14) ˜ ˜ Introducing φ(Ek . (2. we want the most general solution. j) = ↔ i φ∗ ∂0 φ. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. 2. This continuity equation can be written down o covariantly using the components (ρ. k) ≡ a(k) and φ(−Ek . (2. rather than as a wave function.3). however. which is what Lorentz transformations do. there are no longer fundamental objections to mix up space and time. The KG equation. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. −i φ∗ ∇ φ .10) ∂t which follows directly from the Schr¨dinger equation.13) ˜ with (in principle complex) coeﬃcients φ(k). 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.15) . ∂µ j µ = 0. Then. having the KG equation as a space-time symmetric (classical) equation.Relativistic wave equations 9 i. −k) .5). The appropriate current corresponding to the KG equation (see Excercise 2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). This leads to the existence of negative densities.2) is j µ = i φ∗ ∂ µ φ or (ρ. 2M 2M (2. just as the dependence on time was in ordinary quantum mechanics.

18) (2. for which we consider space inversion.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) .e. after restricting the energies to be positive). (2π)3 2Ek (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. Since a · b = a · ˜ it is easy to see that ˜ b. i.6 We will explicitly discuss the example of a discrete symmetry. which implies (xµ ) = (t. (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). φP (x) ≡ φ(˜) (P for parity). changing the sign of the spatial coordinates.14 and 2. Another symmetry is found by complex conjugating the KG equation. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. (2.17) (2. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. (2π)3 2Ek (2. 2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). (2.16) The consequence of this is discussed in Exercise 2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k).23) . x) → (t. 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) . (2π)3 2Ek (2.Relativistic wave equations 10 In Eqs 2. This shows how parity transforms k-modes into −k modes.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) . Performing some Lorentz transformation x → x′ = Λx. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). −x) ≡ (˜µ ). x ∂µ ∂ µ + M 2 φ(˜) = 0.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.

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

i (a) Show that relativistic invariance. Is the time dependence of the result as expected? Note that the mode amplitudes will become creation and annihilation operators after quantization of the ﬁelds (Chapter 7). 2. Note that a{µ bν} indicates symmetrization. . B} = AB + BA. 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. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. Exercise 2.7 To solve the problem with positive and negative energies and get a positive deﬁnite density.Relativistic wave equations 12 Exercise 2. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function.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). j 0 = ψ†ψ and j = ψ † αψ.15. Exercise 2. a{µ bν} ≡ aµ bν + aν bµ . t). ∂t where ψ is a wave function and the nature of α and β is left open for the moment. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . also using the result in (a). β} = 0. We will encounter these quantities later as energy and momentum. i. t) = (−i α · ∇ + m β) ψ(r. 1 2 (b) Express now the full charge QV (t) (exercise 2. in terms of the a(k) and b(k). B] = AB − BA and the anticommutator as {A.e. and β 2 = I.5 (a) As an introduction to parts (b) and (c).1) for a complex scalar ﬁeld current (exercise 2. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). d3 x (∂ {0 φ)∗ (∂ i} φ). (b) Show that one has current conservation for the current 2 We denote the commutator as [A.2) in terms of the a(k) and b(k) using the expansion in Eq. ∂ ψ(r. 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 . {αi . αj } = 2 δ ij I. 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 .

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

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

.e. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. H] = 0 for g ∈ G.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. n) π ≤ ϕ ≤ 2π. thus. . For instance in a function space translations. 2 2 2 (3. i. n) ˆ −→ R(2π − ϕ. The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. Near the identity. 3. one has a Lie algebra isomorphism that is linear and preserves the bilinear product.e.9) The parameter-space of SU (2). φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . for ﬁxed n. it is suﬃcient that the generators commute with H. thus. 2 (3. [g.10) σ ˆ · n.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . the above mapping corresponds to the trivial mapping of the Lie algebras. 2! . . . n) i ∂ϕ = ϕ=0 (3. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I.11) Thus σx /2.13) 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. i. . They satisfy σi σj σk = i ǫijk . also can be considered as a ﬁlled 3-sphere. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. immediately sees that the Lie algebras are identical. n) −→ R(ϕ.12) One. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. in particular that AB → RAB = RA RB ). . however. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. sometimes in more than one way (but this will be discussed later).e. SU (2) ≃ SO(3). (3. n! 2! n=0 ∞ (3. n) ≈ 1 + i ϕ J · n. In the full parameter space. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). H] = 0 for g ∈ G. (3. ˆ ˆ A(ϕ. (3. i. . The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations.14) For a Lie group. Suppose we have a system described by a Hamiltonian H.15) [g.

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

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

23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. We considered the generators and looked for representations in ﬁnite dimensional spaces. consisting of the Lorentz group and translations. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . From this property. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. it is convenient to use a vector notation for the points in Minkowski space. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ .Groups and their representations 18 3. (ii) |Λ00 | ≥ 1.21) g 20 g21 g22 g23 0 0 −1 0 g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. e To derive some of the properties of the Lorentz group. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. (det Λ = +1 is called proper. (3.22) (Note that xT is a row vector).g. G = 10 (3. det Λ = −1 is called improper). σ/2 in a two-dimensional (spin 1/2) case.25) (3.27) (3. In this section we consider the Poincar´ group. 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. Writing x as a column vector and the metric tensor in matrix form. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = .26) (3. e. Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ .3 The Lorentz group In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone .

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

which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. i i0 i i0 1 ijk M jk 2ǫ (3.36) (3. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . K = M ). P+ etc. We therefore ﬁnd (again) that there are six generators for the Lorentz group.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. we continue using covariance arguments. Also this group can be divided into four parts.33) These four transformations form the Vierer group (group of Klein). explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations.39) (e. Summarizing. 3 30 with η = ω and K = M (e. (3. We will just require Poincar´ invariance for the generators themselves. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . There are six generators.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. This is a + normal subgroup and the factor group is the Vierer group. ǫ). three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). a).38) Explicitly.g. The extension to the Poincar´ group is e ↑ straightforward. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. [M µν .35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . Of the four parts only L↑ forms a group.g.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . (3.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. inf (3. 3. In order to ﬁnd the commutation relations including the translation generators. inﬁnitesimally approximated by (I + ω.

For instance. 3. Taking any one of the states in an irreducible representation. P ρ ] = 0. ν (3. P µ |p. that commute e among themselves. [P µ . 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. J j ] = i ǫijk J k . as part of the set. H] = 0. one obtains 2 [P i . For instance.42) (to decide on where the inverse is. Inﬁnitesimally. . . (3.45) Note that the commutation relations among the generators M µν in Eq. writing the generator P = (H/c.46) [K i .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . leads to U (Λ.44) (3. 3. 2 2 (3. K j ] = −i ǫijk J k /c2 . They just state that M µν transforms as a tensor with two Lorentz indices. a) = Λ−1 µ P ν . H] = i P i .Groups and their representations 21 inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. ǫ) = 1 − ωαβ M αβ + i ǫα P α .48) . [J i . = −i (g µρ P ν − g νρ P µ ) .38 could have been obtained in the same way. [K i . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. a) = Λµν P ν or U † (Λ. H] = [J i . (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . P j ] = [P i . The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . The eigenvalues of these will be the four-momentum pµ of the state. P j ] = i δ ij H/c2 . the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators.47) (3. (3. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. We have here reinstated c. P j ] = i ǫijk P k . a)P µ U (Λ. . = pµ |p. . Explicitly. (3. the generators M αβ no longer exclusively act in Minkowski space. P ν ] [M µν . the generators J 2 and J 3 in the case of the rotation group.41) 2 At this point. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. the generators J± in the case of the rotation group. [J i . [K i . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . a)P µ U † (Λ. K j ] = i ǫijk K k . known from non-relativistic quantum mechanics are obtained. P ) in terms of the Hamiltonian and the three-momentum operators. These form a group called the little group associated with that four vector. . The generators of the translations are the (momentum) operators P µ . (3. . other states in that representation are obtained by the action of operators outside the maximal commuting set. [J i .

Wν ] = −i (gµα Wν − gνα Wµ ) . 0. = = i exp − ǫµνρσ M µν pρ sσ |p. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . which can thus be chosen spacelike. . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. . . while the others have no physical signiﬁcance. . Massive particles: p2 = M 2 > 0.52) (3. the third case represents the vacuum. Wα ] = [Wµ . i ǫµνρσ W ρ P σ . (3.49) (3.51) The following properties follow from the fact that W µ is a four-vector by construction. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. 0. . 2 (3. . 2 exp (−isµ W µ ) |p. M ǫijk . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). (3.56) . 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.54) This will enable us to pick a suitable commuting ’spin’ operator. p0 > 0. . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . .53) (3.55) commute with all generators and therefore are invariants under Poincar´ transformations. Pν ] = [Wµ .

(3. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). 0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3.57 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.64) such that in an arbitrary frame where the four vectors p.e. W j (p)] = i M ǫijk W k (p). .58) i. M2 (3. 2 . show that W /M can be interpreted as the intrinsic spin. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 0. 0. (0.e. The commutation relations [W i (p). . M J) with components W i = (M/2)ǫijk M jk . 0.65) . 1.59) (3. (3. 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. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p).Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. 1. 0). 1.57) (i. 1 2 (3.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. 0.61) one sees that S i (p) i =− W2 . 0). In that case the vectors ni (p) are just the space directions and W = (0. p. 0. n1 (p). W i (p) = −W · ni (p)). |p3 |). Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).. s. Together with the four momentum states thus can be labeled as |M. i (3. (1. 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. M2 (3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. Having made a covariant decomposition. (0. 0). p0 > 0. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . . ms . 3.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 .

W 2 (p)] [λ(p).70) as a commuting set with zero eigenvalues for P 2 and W 2 .72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. Thus we have P 2 . |p. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). which can take any integer or half-integer value.71) which is called the helicity.68) (3. 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). −i W 1 (p). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises).Groups and their representations The above does actually include the massive case. the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a . W 2 and λ is derived from the general commutation relations for [Wµ . P µ . W µ (p) = λ(p) P µ . a)|0 = |0 . Thus 2 λ(p) = J ·p . where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2).73) (3. A massless particle. i W 2 (p). (3. But if M = 0 one has a diﬀerent algebra (Eq. Note that angular momentum does not contribute to helicity as L · p = 0. The vacuum: pµ = 0. Exercises Exercise 3.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). is characterized by its momentum and the helicity.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). thus.1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. 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).13. |p| (3. W 2 (3. This state is in physical applications nondegenerate and is denoted by |0 . U (Λ. (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. and gives [W 1 (p). 1 2 24 (3. W 1 (p)] [λ(p). Wν ]. however. 3. W 2 (p)] = = = i M 2 λ(p). The algebra of W 1 . implying continuous values for the spin (actually for W i = M S i ). The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. We don’t know of any physical relevance.67 has a vanishing right hand side). λ .67) (3. 3.

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

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

B j ] = i ǫijk Ak .4) (4.6) (4. 0. The matrices in SL(2. C) is the group of complex 2 × 2 matrices with determinant one. K = +i .2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). This tells us how to ﬁnd the representations of the group.1 The Lorentz group and SL(2. They will be labeled by two angular momenta corresponding to the A and B groups. respectively. becoming the deﬁning representations of SL(2. (4. (j. 2 2 σ σ J = . 2 (4.7) From the considerations above.1) (4. 0) Type II : (0. C).9) (4. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. It is simply connected and forms the covering group of L↑ . They precisely correspond to the two types of representations that we have seen before. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . j ′ ). 2 1 (J − i K). i ǫijk B k . Special cases are the following representations: Type I : (j. U (ϕ) H(η) (4. K = i J (A = 0). B j ] = [Ai .10) . Aj ] = [B i . 2 2 (4. C) Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (hermitean) combinations A = B which satisfy the commutation relations [Ai . where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞.Chapter 4 The Dirac equation 4.3) (4. K = −i . The group SL(2.5) = 1 (J + i K). C) can be written as a product of a unitary + matrix U and a hermitean matrix H. similarly as the homeomorphism between SO(3) and SU (2). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . j) K = −i J (B = 0). (4.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. With this choice of parameter-spaces the plus and minus signs are actually relevant.

U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . This shows that M ∗ ≃ M .1). The Lorentz transformations Λ(M ) and Λ(M ). C). however. U σ µ aµ U −1 = σ µ ΛR aµ .The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. C). thus.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. 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. ξ −→ Hξ. σ µ ≡ (1.12) ξ → U ξ. such that M = SM S −1 . 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. there exists the matrix ǫ = iσ 2 . Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). if there does not exist a unitary matrix S. One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. This is not true for the two-component spinors in SL(2.16) = σ µ Λ R aµ .15) (4. In fact. while ±ǫη ∗ transforms as a type-I + spinor. C) or L↑ .18) As already discussed for SU (2). thus. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. ˜ = σ µ Λ B aµ . (4. (ǫ∗ = ǫ. (4. Λ(ǫ) ∈ L+ . Eqs 4. H(η) = exp(−η · σ/2). η → U η.e.13) 2M (E + M ) (exercise 4. (4. one can use a boost to the frame with momentum ˆ ˆ p.11) Considering ξ and η as spin states in the rest-frame.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. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. H ≡ (H † )−1 = H −1 ). C) and. The following relations for + rotations and boosts are useful. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . ˜ (4. ǫ−1 = ǫ† = ǫT = −ǫ). (4. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. (4. σ).19) . ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. η → Hη. −σ). cannot be connected by a Lorentz transformation belonging to L↑ . i. namely H(η) = exp(η · σ/2). 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).

0) of SL(2. a) and ˜ a = (a0 . −→ M (Λ) (4. therefore. 1 ) and ( 1 . 1 ) and ( 1 . (pseudoscalar).e. 0) are inequivalent. we obtain for the two components of u which we will refer to as chiral right and chiral left components. 0 χm uR H(p) . From the deﬁnition of the representations (0. A parity transformation changes aµ into aµ . the well-known two-component spinor for a spin 1/2 particle.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. ǫµνρσ will change sign.20) 4. = (4. generators. m) = χm . i.g. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. m) = η(0. but rather the operators W i (p) should be used. It has the same eﬀect as lowering the indices. as we have seen in section 4.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. The behavior is the same for classical quantities. C). e.13. −→ −r (vector). but to L↑ .23) u(p. −a). We also have seen that the representations (0. m) = χm uL H(p) 0 . (scalar). the boost in Eq. 2 ) and ( 2 . Thus M ∗ and M are not equivalent within SL(2. C) are suitable for representing spin 1/2 particles. i. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . (axial vector).2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0.e. they can be connected.1. (4. the aij elements of a tensor will not change sign. ) −→ ( . Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). For the spin 1/2 representations of the Poincar´ group including parity we. that have the correct commutation relations. Although these operators cannot be used to label the representations. 4.17. The representing member of the class P− is the parity or space inversion operator Is . and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. must combine the repe resentations. Within the Lorentz group. (vector). we have seen that in the rest frame W i (p)/M → J i . This provides the easiest way of seeing ˜ what is happening. they cannot be connected by a unitary 2 2 transformation. −→ −p (vector). Taking M (Λ) = H(p). but by a transformation belonging ↑ ↑ to the class P− . The i angular momentum operators J are represented by σ i /2. 4. etc. 0). For a particle at rest only angular momentum is important and we can choose ξ(0. 2 2 (4.The Dirac equation 29 where The matrix I2 does not belong to L↑ . where a = (a0 .

This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. 0 H 2 (p) uR uR . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. The general requirements for the γ matrices are thus easily obtained.29) which is an explicit realization of the (momentum space) Dirac equation. (iγ µ ∂µ − M ) ψ(x) = 0. 2 (4. which in general is a linear equation in pµ . 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. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). . (4. γ } ∂µ ∂ν + M 2 ψ(x) = 0.32) suppressing on the RHS the identity matrix in Dirac space.33) deﬁne for a four vector a the contraction a = aµ γµ / . (4.28) which is a covariant (linear) ﬁrst order diﬀerential equation.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . It is of a form that we also played with in Exercise 2. The explicit realization appearing in Eq. From this one obtains the Cliﬀord algebra for the Dirac matrices. = (4.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. We will discuss another explicit realization of this algebra in the next section.7. (4. The ﬁrst indeed is solved. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. = H −1 (p).24) (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). {γ µ .27 is known as the Weyl representation. (4. (4. γ ν } = 2 g µν . 4. this can be rewritten 1 µ ν {γ .30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. p (4. Since ∂µ ∂ν is symmetric.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .

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. Relying on the Pauli exclusion principle for spin 1/2 particles. (4.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices.35) is indeed positive and j 0 can serve as a probability density. . The density j 0 = ψγ 0 ψ = ψ † ψ (4.3 General representations of γ matrices and Dirac spinors {γ µ .34) for the adjoint spinor ψ ≡ ψ † γ0 . From the vacuum a particle can be removed. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. . (4. the one of the negative energy states. This can most easily be seen in the particle rest frame. E = +M (twice) and E = −M (twice). From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).g. remains. The Dirac sea forms the vacuum.1: The Dirac sea of negative energy states and antiparticles. The second problem encountered before. This hole forms an antiparticle with the same mass. 4. γ ν } = 2 g µν .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 .36) (4. (4. but with properties such that it can be annihilated by the particle (e.27).37) Using the explicit form of γ in the Weyl representation (see Eq. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. 4. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. This problem was solved by Dirac through the introduction of a negative energy sea. one sees that there are two positive and two negative eigenvalues. opposite charge).The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.

ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. For example.45) (γ5 )W.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.R.R.R. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . while L−1 γ µ L = Λµ γ ν . (4.3) .40) (4. 0 (4. γ µ } = 0 and explicitly one has 0 1 1 . γ 2 ].41) (4. S −1 . 4! (4. γ σ ]. We note that ψ → Lψ and ψ → ψ L−1 .46) (4.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . 2 (4.43) S=√ . For instance in the Weyl representation (but valid generally). the rotation and boost generators in Weyl representation in Eqs 4.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . is 1 1 1 (4. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .R.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . (γ5 )S. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. −→ Sγµ S −1 . Another matrix which is often used is γ5 deﬁned as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisﬁes {γ5 . it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. (γµ )S. −→ Sψ. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .9 and 4. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k .10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . γ 0 ]. = S(γµ )W. =ρ ⊗1= 0 −1 3 (4. which in the case of the Weyl representation are just the upper and lower components.48) are projection operators. that project out chiral right/left states.47) (4. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.

58) where ηc is an arbitrary (unobservable) phase.R. 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) = 0.56) (i/ − M ) ψ c (x) = 0.6). x) → x = (t. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . Some properties of C are C −1 = C † and C T = −C. reverse the charge of the particle. for instance. that brings ψ → ψ c . but it does. e. usually to be taken unity. (i/ − M ) ψ(x) = 0. = 0 ǫ iσ 2 (4.g. C is real and one has C −1 = C † = C T = −C c and [C. which is again a solution of the Dirac equation. 0 −ǫ 0 0 −ǫ 0 .50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. T (4.52) η ξ one can apply space inversion. is a left-handed spinor.54) e. The latter implies that the conjugate of a right-handed spinor.57) (4. ψ= (4. γ5 ] = 0. ∂ with the solution ψ c (x) = ηc Cψ (x).g. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. such that T Cγµ C −1 = −γµ . In S.R. In any representation a matrix C exist. −iσ 2 0 −ǫ = . in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . parity. called charge conjugation. One has ψ c = −ψ T C −1 .59) −→ ψ c = C ψ = η ǫ ξ∗ .R. ψR . ψ= (4.) and all representations connected via a real (up to an overall phase) matrix S. This symmetry does not change the spin 1/2 nature. x = (t. Explicitly. (C)W.R.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. (or W. ∂ (4.55) (4. = = 0 iγ γ = −iρ ⊗ σ = −iσ 2 −iσ 2 iγ 2 γ 0 = −iρ3 ⊗ σ 2 = 0 2 0 1 2 Thus we ﬁnd back the Dirac equation.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. T (4. As with parity we look for a transformation. ∂ (4.53) (4. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. (C)S.

ψ(x) = v ±s (p) ei p·x . Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). p (4. the spin polarization vector in the rest frame is the starting ˆ point. s)v(p. s′ ) = 2Ep δss′ . s) = u(p. s)¯(p. s) = v(p. the best representation to describe particles at rest is the standard representation. v(p.61) σ·p −(Ep + M ) There are also two negative energy solutions. (2π)3 2Ek (4.62) where χs are two independent (s = ±) two-component spinors. s) = Ep + M σ ·p Ep +M χs .. v (p. s) ei k·x d∗ (k. In that frame is a spacelike unit vector sµ = (0.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) = s d3 k u(k. s) and C v T (p. s ) = v (p.70) = Ps = ˆ 0 1∓σ·s 2 2 . where u satisﬁes −σ · p Ep − M u(p) = 0. s) . be obtained by a Lorentz transformation.64) (4. the spinors satisfy C uT (p. with E = Ep = + p2 + M 2 . s)v(p. s)v(p. ⇔ (/ + M ) v(p) = 0. ¯ ¯ † ′ u (p.63) u(p.1). ⇔ (/ − M ) u(p) = 0. s) u = . s)¯(p. ∂ −iσ · ∇ −i ∂t − M (4. s) + v(k.g. s) = u(p. ¯ (4. (4. s). The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle.68) (4.69) = − s In order to project out spin states. ψ(x) = u±s (p) e−i p·x .66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. ¯ ′ ′ Explicit solutions in the standard representation are χs . s) e−i k·x b(k. 2M 2M −/ + M p v(p. where v satisﬁes σ·p −(Ep + M ) v(p) = 0.65) (4. s)u(p. 2M 2M (4. In an arbitrary frame one has s · p = 0 and s(p) can e. in which γ 0 is diagonal (see discussion of negative energy states in section 4. s)u(p. s)u(p.The Dirac equation 34 4.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors.60) ψ(x) = 0. The ¯ ¯ solutions are normalized to u(p. s) v = . s′ ) = 2M δss′ . s′ ) = −2M δss′ . s). s ) = v † (p. s ) = 0.

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

s)(γ ν )kl u u s. Some elementary properties are: Properties of products of γ-matrices: (1) γ µ γµ = 4 (2) γ µ γ ν γµ = −2 γ ν or γ µ /γµ = −2 a. however. (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . s′ )) (¯(k.75) Lµν (k. In principle one can take a representation and just calculate the quantity u(k. namely the calculation of the quantity 1 ∗ (¯(k. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ .g µρ g νσ + g µσ g νρ ). (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. k ′ ) = 2 ′ s. ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ .s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . s)γ µ u(k ′ . u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ . s′ )(γ µ )ij uj (k. s)(γ ν )kl ul (k ′ . (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. For this we ﬁrst have to prove (do this) that (¯(k. s)¯k (k. s′ ). u u (4. s)γ µ u(k ′ . s′ )(γ µ )ij uj (k. s)γ ν u(k ′ .76) ul (k ′ .s′ ∗ (4.5 γ gymnastics and applications An overview of properties can be found in many text books. s′ ) ¯ u s. (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ).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(k ′ . s′ )) .) In order to show the use of the above relations. (6) Tr(γ5 ) = 0. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . s).) (2) Tr(1) = 4.The Dirac equation 36 4. etc. γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. s)¯k (k. more convenient ¯ to use the projection operators introduced earlier. s′ )γ µ u(k. (Use (γ5 )2 = 1 and pull one γ5 through. (8) σµν ≡ (i/2)[γµ . a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b. s′ )¯i (k ′ . k k 2 . (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 .g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . (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 ′ . we will give one example. (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. It is.

γ σ ] = σ ρσ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M.4 Show that in P+ ≡ s=±1/2 u(p. s) and v(p. (4. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. a 4 a · b. x ˜ . i S ρσ ] = g ρµ γ σ − g σµ γ ρ . s)¯(p.g. x → x. H(p) = exp η·σ 2 = M +E+σ·p .The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. where x = (t. 4 2 Exercise 4. Do this e. Exercise 4. Show that the equality holds. 0 if n is odd. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ .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.g. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). by letting them act on the four basis spinors u(p. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . 1 i −i S ρσ = − [γ ρ . s).1 Prove Eq. s) u p /+M = 2M 2M both sides are projection operators. e. −x) is also a solution of the Dirac equation.13. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . Excercise 4. ˜ Exercise 4.77) Exercises Excercise 4. 4.5 Show starting from the relation {γµ . of which only the traces of four and two gamma-matrices are nonzero.

M R = 1. Note: To prove this. i. is a stationary solution of the (free) Dirac equation.5 and M → ∞. Note: use ﬁrst the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions.66 MeV. Calculate the value of k for M R = 0. n It turns out that there is actually a whole family of conditions that provide conﬁnement. Excercise 4. For such a cavity nµ = (0. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless.57 MeV. Calculate for now just |M |2 . mµ = 105.The Dirac equation 38 Exercise 4. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. r). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. . namely i / ψ = eiαγ5 ψ.8 × 10−6 . you need to determine the corresponding conﬁnement condition for ψ. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . the decay would even be forbidden. 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 ). we want to conﬁne this solution to a spherical cavity with radius R. me = 0. there is no current ﬂowing through the surface of the cavity. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal).e. The pion has spin 0. We will not pursue this chiral freedom at this point and take α = 0.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). ν p Note: how such parametrizations work will be explained in chapter 8. Do you understand why? The calculation of the life-time is completed in chapter 11. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0.511 MeV).7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. 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. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. (b) Next.

These are solutions of the Klein-Gordon equation with in addition a condition.6) where Aµ is the four-vector potential (5.7) . In the restframe.5) −E B2 . with ∂µ V µ = 0 (Lorentz condition). λ) + ǫ(λ)∗ (k) eik·x c∗ (k. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0.2 The electromagnetic ﬁeld F µν = = ∂ µ Aν − ∂ ν Aµ 1 −E 2 0 −E 1 E 0 −B 3 2 E 3 B 0 3 E −B 2 B 1 Aµ = (φ.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.Chapter 5 Vector ﬁelds and Maxwell equations 5. 39 (5.2) (5. λ) . These can describe spin 1 states in the rest-frame. where k = k0 = (M. µ (2π)3 2E µ (5. (2 + M 2 )Vµ = 0. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. (5.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. The vectorﬁeld is therefore (λ) suited to describe spin 1. ǫλ ). 0). just as the two-component Dirac spinors did for spin 1/2. M2 (5. ±(1) are obtained by boosting the restframe vectors. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . there are then indeed three independent possibilities. A). namely ǫµ (k0 ) = (0.1) The Lorentz condition implies that k µ ǫµ (k) = 0.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.1 Fields for spin 1 In section 3.4) λ=0. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . In an arbitrary frame ǫµ (k) with λ = 0.± 5.

j) the Maxwell equations read ∂µ F µν = j ν . . The equation in vacuum. AP (x) = Aµ (˜).18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ.10) the electric and magnetic ﬁelds are unchanged.11) This equation is not aﬀected by a gauge transformation. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. in which case one has 2Aµ = jµ . 40 (5. if in Eq. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. moree over. 0. Furthermore. transverse or Coulomb gauge). 2Aµ = 0. (5. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). only two independent vectors remain ǫ(±) µ (k) = (0. in accordance with what we found in section 3.12) (5. Taking the divergence of this equiation. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. Therefore if k µ = (|k|. For electrodynamics one has furthermore the freedom of gauge transformations. (5. ∓1.15) (5.13) Hence. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x).17) (radiation. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ .9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. 0. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. |k|). (5. (5. As remarked above. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . 5. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld.8) (5. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. −i. Although the Lorenz condition is a constraint.16) (5. i.5.15 the d’Alembertian is replaced by 2 + M 2 . for M = 0. if ∂µ Aµ = 0 for large times |t|. ∂µ Aµ = 0.e.14) of which the solutions give the Li´nard-Wiechert potentials. 0). ˜ x Note that via parity transformation one obtains another solution. It states the absence of magnetic charges and Faraday’s law. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. it will vanish at all times. One possible choice is the gauge A0 = 0 (5. (5.

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

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

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

and the total change x′ (t′ ) = x(t) + ∆x(t). ∂x ˙ (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 . Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current.Chapter 6 Classical lagrangian ﬁeld theory 6.5) with ∆x(t) = δx(t) + x(t) δτ . x (t) = x(t) + δx(t). As an example. recall classical mechanics with the action t2 S= t1 dt L(x.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. ˙ 2 (6. thus t′ = t + δτ.3) (6. x) = K − V = ˙ 1 mx2 − V (x). (6. These equations can be obtained from a lagrangian using the action principle. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.8) 44 .6) The ﬁrst term leads to the Euler-Lagrange equations. x) ˙ (6. t1 (6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). ∂x (6. ˙ p= ∂L . d dt ∂L ∂x ˙ = ∂L .1) and as an example the lagrangian L(x.

x) ≡ p x − L . − d3 x . ∂µ φ′ . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). (R3 . (6. Consider a lagrangian density L which depends on these functions. t1 (6. H is a conserved quantity. δ(∂µ φ) (6. dσµ .18) 1 Taking a functional derivative. . ˙ x The second term in Eq. In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). t1 ) and R3 .t2 ) d3 x .13) (6. We will come back to it in a bit more formal way in section 9. The hamiltonian H is deﬁned by H(p. . = φ(x) + δφ(x) (6. R (6.17) Furthermore the variations δφ and δ∂µ φ contribute to δS.6 can be rewritten as δS = . . but which in 3 dimensions includes conserved quantities related to rotations and boosts.t1 ) (6. ∂µ φ.14) (6. ∂µ φ(x)). ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . x′ ) − d4 x L (φ.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). 6. . ˙ (6.e.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 . λ=0 (6. = ∂R (R3 .12) Here R indicates a space-time volume bounded by (R3 . with ∆φ(x) = δφ(x) + (∂µ φ)δx . i.Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. ∂µ φ(x)) = R d4 x L (φ(x). indicated with δF [φ]/δφ should pose no problems. + ∂L ∆x − H δτ ∂x ˙ t2 . . In general any continuous transformation. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ.15) d4 x′ L (φ′ . x). . The resulting variation of the action is δS = R µ ′ = xµ + δxµ . their derivatives and possibly on the position. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). . t2 ).2. which in classical mechanics vanishes at the boundary) does not play a role. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . L (φ(x).16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . . ∂µ φ(x). . .11) λ=0 of which p and H are the simplest examples related to space and time translations. Variations in the action can come from the coordinates or the ﬁelds.

For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. (6. Using the variations in ψ (in the symmetric form). 2 ∗ (6.27) (6. (6. the second term is irrelevant. (6.19) δ(∂µ φ) δφ 6. leading to (2 + M 2 )φ(x) = 0. 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). The integrand of the ﬁrst term must vanish.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration). leading to the Euler-Lagrange equations. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.23) (6. One easily obtains (2 + M 2 )φ(x) = 0. 2 (6.25) (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.2 Lagrangians for spin 0. 2 ψ = 0.20) (6.24) (6. ∂ (6.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.26) The Dirac ﬁeld The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities.29) . the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. (2 + M )φ (x) = 0.21) which diﬀer only by surface terms. δL δL ∂µ = .

L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .g.38) (6. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. and thus (β ∗ α)∗ = α∗ β = −βα∗ . ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . which comes because we in essence use ’real’ spinors. ∂ ∂ 2 2 (6. . The Majorana case is in fact more general. 2 (6. 1 M η † ǫη ∗ − M ∗ η T ǫη .32) There exists another possibility to write down a mass term with only one kind of ﬁelds.36) (6. namely2 LM (Majorana) = + (6. it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left ﬁelds ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors. ψL ≡ (ψL )c = C ψL = 0 (6.and lefthanded ones. (6.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . Using the twospinors ξ and η. 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.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .31) showing e.7). The reasons for Grassmann variables will become clear in the next chapter.34) (6. that the lagrangian separates into two independent parts for M = 0. Peshkin and Schroeder or Exercise 12.g. the mass term in the Dirac lagrangian 6. (6.31 is given by LM (Dirac) = −M ξ † η + η † ξ . 2 αβ = −βα. (6. but note the factor 1/2 as compared to the Dirac lagrangian.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . or equivalently separate the ﬁeld into right. Vector ﬁeld From the lagrangian density L 1 1 λ 2 = − Fµν F µν + M 2 Vµ V µ − (∂µ V µ ) 4 2 2 1 µ = V (∂ 2 + M 2 )gµν − (1 − λ)∂µ ∂ν V ν . We note that of which the left part coincides .

(6.. t). δ(∂µ φ) For continuous transformations. ∂µ J µ (x) = 0.σ1 with σ = (R3 .44) δL ∆φ + Θµν (x)δxν . the presence of a symmetry that leaves the lagrangian invariant.18. (6. Q= d3 x J 0 (x. 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. + ∂R dσµ dσµ ∂R ≡ where d x . . discussed in the next chapter. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. . (6. (6. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e.3 Symmetries and conserved (Noether) currents Next.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x).42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . These do not aﬀect the dynamics and will give rise to conserved currents. which in a consistent picture precisely generate the symmetries. Q(t1 ) = Q(t2 ).45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .41) (6. these conserved quantities become operators. requires the presence of a conserved quantity. In this case.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. δ(∂µ φ) (6. (6. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . 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 . δ(∂µ φ) (6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). 6.40) 6.. we consider the second term in Eq.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 . As an example for the classical ﬁeld. In particular when the lagrangian is invariant under symmetries.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .47) . t). one can consider the variations at the initial or ﬁnal surface.

Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . (6. 6.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.50) (6.53) These are the energy and momentum. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld. .4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. (6. e. 2 (6.52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0. From Noether’s theorem one sees that the current Θµν ǫν is conserved. In the next section we will see the quantity show up as the charge operator. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. (6.49) These currents are conserved as discussed already in chapter 2.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.54) (6. we obtain (since δxµ = 0).g. Summarizing.51) (6. including e the space-time translations and the Lorentz transformations.3) j µ = i e φ∗ ∂µ φ.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . t) is the conserved charge (Q = 0).55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). = 0. The integral over the zero-component. H = d3 x Θ00 (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 . 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. ˙ Q = d3 x j 0 (x. ↔ (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.63) µν = ∂µ Θ . where H ρµν is the quantity appearing in M ρµν .58) (6. however. independent of x.67) i. φ(x). Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). 2 (6. In general this tensor is not symmetric.59) (6.3 we have seen global transformations or gauge transformations of the ﬁrst kind. (6. e. in which the U (1) transformation involves an angle e Λ. (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). ∂µ φ(x) + i e ∂µ Λ(x) e (6.69) i eΛ(x) φ (x). In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. the angle of the transformation depends on the space-time point x. is not invariant under local gauge transformations. ∂µ T µν (6. Any lagrangian containing derivatives.65) M µρσ = T µρ xσ − T µσ xρ . −i eΛ(x) ∗ i eΛ(x) (6.3).68) (6. (6. • Vector ﬁeld Aµ : −i Sρσ = aρσ . • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . 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 . (6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.60) 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 .6) T µν = T νµ . 6.66) In section 6. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. Deﬁning T µν = Θµν − ∂ρ Gρµν .e. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).61) A ﬁnal note concernes the symmetry properties of Θµν . one has a tensor that satisﬁes (exercise 6.64) (6.70) . (6.g.

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) − Fµν F µν . (6.72) (6.76) 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 . ∂µ φ) =⇒ L (φ.78) . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).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 . 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 . 1 L (φ. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. −e j µ Aµ = −e ρφ + e j · A.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. ∂ 2 4 (6. 4 (6. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x).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).Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. 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. (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . (6. D ∂ A (6.73) (6. As an example consider the Dirac lagrangian. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.

∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). −eψγ ν ψ. giving the Maxwell equation coupling to the electromagnetic current.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . φ → eiΛ φ. and deduce what is the charge of the antiparticle as compared to a particle. ∂µ F µν = j ν . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. i. Exercise 6.1 (a) Show that the Klein-Gordon equation for the real scalar ﬁeld can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 .Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν .79) Exercises Exercise 6.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds. ∂µ F µν = j ν . . ↔ Exercise 6. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. Exercise 6.4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. where j µ = e ψγ µ ψ.e. 2 2 (b) Show that the Klein-Gordon equation for the complex scalar ﬁeld (considering φ and φ∗ as independent ﬁelds can be derived from the lagrangian density L = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ. (6. give the equation which is satisﬁed by ψ c = Cψ . 4 (b) What is the form of the interaction term involving a current jµ and the ﬁeld Aµ that will give the inhomogeneous Maxwell equations.

e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r.6 (optional) Prove the properties of the tensor T µν in section 6. A) (see also exercise 6. .4. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . 2me Exercise 6.4) 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 . t). Comment: this relation is known as the Pauli equation. It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic ﬁeld takes the value gs = 2.Classical lagrangian ﬁeld theory 53 Exercise 6. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. t) = Enr ψu (r.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 . 2M 2M with Enr ≡ E − M ≪ M . in order to show that Θρσ − Θσρ = ∂µ H µρσ . use ∂µ T µν = 0 and ∂µ M µρσ = 0. Hint ≡ −gs µe · B = −gs Qe s · B. This relation is known as the Pauli equation.

[q. The bilinear product for the conserved quantities. (7.3) For variations δO(p. (7. B]P = dB dA dA dB − . Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. t).2) H(p.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t).5) ∆ x ∆3 x 54 . there are conserved ˙ ˙ quantities Q (Eq. p]P = 1.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q.Chapter 7 Quantization of ﬁelds 7. (7. is the Poisson bracket [A. Q]P . q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. however.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. 1 qx (t) = 3 d3 x φ(x. p]P .11) found through Noether’s theorem. Further. q) also considered in the previous chapter. starting from the lagrangian L(q. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. dq dp dq dp (7. in our example p = mq. q) = ˙ 1 mq 2 − V (q). Quantization of the theory is achieved by imposing canonical commutation relations between q and p. p] = i. ˙ 2 (7. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. dλ Examples are dO = [O. ˙ L(q. labeled by the position. namely [q. 6. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. q(q. q) = p q − L(q. H]P dt and dO = [O.

• Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). (7. ∂ qx ˙ ∂ qx ˙ (7. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . i U † (Λ. (7. t)] = [Π(x. t).10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.15) with furthermore the relations [φ(x. t)] = 0. ∆φ. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). t). for the scalar ﬁeld theory. discussed in the section 6. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . t)] = i δ 3 (x − y). ˙ δ φ(x) (7. for symmetry transformations. qx+∆x ) . ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ).2.12) (7. we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. a) φi (x)U (Λ. Sometimes it is useful to keep in mind that. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). Π(y.6) ∆3 x L x (qx . albeit with ’sharp’ functions. φ(y. • Poincar´ invariance e The full variation of the ﬁelds. formally. As an example. ∇φ) (7.14) In fact.8) (7.11) (7. qx . the discretization procedure above is an explicit example. The essential conditions to consistently quantize a theory then are the following.16) . a) = Rj (Λ−1 ) φj (Λ−1 x − a). φ. Φ(f ) = d4 x φ(x) f (x). (7.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . Π(y. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x.7) d3 x L (φ(x). ∂µ φ(x)) = ˙ d3 x L (φ. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. 2 2 2 (7. t).Quantization of ﬁelds 55 etc.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta.

P ] 2 (7.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. P ] = i. a] = [a† .4.19) 7.23) The hamiltonian in this case can be expressed in the number operator N = a† a. The measurements cannot inﬂuence each other or [Φ(f ). (7. sometimes referred to as second quantization.18) with S µν given in 6. a† ] = a† . a† ] = 0. [Q. [N. [φ(x). a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . (7. j 2 one has [φi (x).Quantization of ﬁelds 56 or if U (Λ. Q] = [P. 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. ω a† a + 2 2 ω i − [Q. Φ(g)] = 0.2 Creation and annihilation operators Before discussing (real and complex) scalar ﬁelds and Dirac ﬁelds we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. 2 i ωµν (S µν )i . a] = −a. Microscopic causality implies local commutativity.25) . φ(y)] = 0 if (x − y)2 < 0. a† ] = 1.24) It is straightforward to ﬁnd the commutation relations between N and a and a† . H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . i(xµ ∂ν − xν ∂µ )φ (x) + i (7. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q .17) (Sµν )i φj (x). Mµν ] = i i∂µ φi (x). j (7. (7. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. Pµ ] = [φ (x). [a. 2 2 (7. and [N. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). P ] = 0 (7.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a.

26) a(k) e−i k·x + a† (k) ei k·x . † √ a and a act as raising and lowering operators.g. 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 .4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). e. 7. Note that we will often write a(k) or a† (k). φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . (7.like’ commutation relations between a(k) and a† (k ′ ). i.e.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).e. Π(x )]x0 =x′0 = 0.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n.33) . (n − 1) a|n † i. (2π)3 2Ek (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .28) (7.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. a† (k ′ )] = 0. the real scalar ﬁeld φ(x) becomes d3 k (7. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). which represents the ’number of particles’ with momentum k. φ(x )]x0 =x′0 = [Π(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. is equivalent with [a(k). The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. a(k ′ )] = [a† (k). and (7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator .3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. [φ(x).30) [φ(x).32) (7. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . and we see that a state |0 must exist for which N |0 = a|0 = 0. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .

For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. deﬁning particle states |k = |k (with positive energy. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . (2π)3 2Ek d3 k a† (k) ei k·x . . (2π)3 2Ek (7. . is solved by subtracting it as an (inﬁnite) constant. d3 k a(k) e−i k·x . i. writing all annihilation operators to the right of the creation operators. |0 .41) (7.34) where V = (2π)3 δ 3 (0) is the space-volume.35) where the vacuum contribution disappears because of rotational symmetry. The rest of the states are then obtained from the groundstate via the creation operator. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).44) p|φ− (x)|0 = ei p·x . (2π)3 2Ek (7. n1 ! n2 ! (7. .40) This procedure is known as normal ordering.e.45) (7.46) (7. |k = a† (k)|0 . which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). k 0 = Ek = k2 + M 2 ).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. (2π)3 (7.43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y). which now contains an inﬁnite number of oscillators. . and multiparticle states |(k1 )n1 (k2 )n2 . d3 k |k k|.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. 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).37) (7.39) The problem with the vacuum or zero-point energy. a(k)|0 = 0.42) (7.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). . assuring that the vacuum (by deﬁnition) has eigenvalue 0!. (2π)3 2Ek (7. (2π)3 2Ek (7. The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. 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). This term will be adressed below.

(7. φ− (y)] ≡ i∆+ (x − y).56) = −δ 3 (x).50) or as integrals over d4 k. d3 k ei k·(x−y) = −i∆+ (y − x). 2π (7. = − (2π)3 2Ek (7. φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).49) (7.52) The result for the invariant commutator function is [φ(x). i.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . The last item to be checked for the scalar ﬁeld are the causality condition. In order to calculate [φ(x). x) = 0 and hence ∆(x) = 0 for x2 < 0. (iii) ∆(0. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .47) ′ d3 k ′ e−i k·x+i k ·y [a(k).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.e. ∆(x) = − ǫ(x0 ) δ(x2 ). (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. φ+ (y)] d3 k e−i k·(x−y) .18). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. d3 k = (2π)3 2Ek = (7. (2π)4 (7.17) and (7.48) (7. (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. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (7.51) (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ(y)] consider µν [φ+ (x). that they satisfy the required commutation relations in Eqs (7.53) t=0 . (2π)3 2Ek ≡ i∆− (x − y).

(2π)3 2Ek (7. which is equivalent to the relations (only nonzero ones) [a(k). φ† (y)] = i∆(x − y). 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) .60) = i φ∗ ∂µ φ.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. (2π)3 2Ek (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (2π)3 2Ek (7.57) (7.e.62) (7. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). we will consider it as a repetition of the quantization procedure.63) i.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x).66) .65) 7. = ∂{µ φ∗ ∂ν} φ − L gµν . ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). a† (k ′ )] = [b(k).61) (7.59) (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) . extending it with the charge operator and the introduction of particle and antiparticle operators.Quantization of ﬁelds 60 7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . (7. 2 From the lagrangian density (7. ↔ (7.

s)¯(k.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). i. (7. s)v(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) .e.74) (7.75) (7. (7. † {ψi (x). (2π)3 2Ek (7.78) (7. s) ei k·(x−y) x0 =y 0 . (2π)3 2Ek d3 k b† (k.71) (7. s)u† (k.70) where the last line is obtained by using the Dirac equation. jµ Θµν = = ψγµ ψ.73) (7. s′ )} = {d(k. (2π)3 2Ek (7. s)ei k·x .76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. s)u(k. s). s) e−i k·(x−y) j s † + vi (k. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s)e−i k·x + d† (k. u v (2π)3 2Ek (7. ↔ i ψγµ ∂ν ψ − 2 (7. . s)e−i k·x .67) i ψ / ψ − M ψψ gµν . s)¯(k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .69) (7. The quantized ﬁelds are written ψ(x) = s d3 k b(k. d† (k ′ . interchanging creation and annihilation operators.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. s).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) : . ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. b† (k ′ . ∂ 2 ↔ (7.77) Note that achieving normal ordering. s)ei k·x + d(k. s)vj (k. The solution is the introduction of anticommutation relations. {b(k.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered.

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). ψ j (y)] = (i/x + M )ij i∆1 (x − y). ˙ δ Ai (7. ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y).82) where i∆1 = i∆+ −i∆− .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.84) (7. −λ(∂ρ Aρ ). For the scalar combination {ψi (x). it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. ψ j (y)} at arbitrary times one has {ψi (x).86) (7. but has the problem of being noncovariant. one has † {ψi (x). we would have obtained [ψi (x). (7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .6 The electromagnetic ﬁeld 1 L = − Fµν F µν . 4 2 This gives the equations of motion discussed before.80) 3 = δ 3 (x − y) δij . ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. (7.88) .Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4.83) From the lagrangian density the canonical momenta are = = (7. 4 Π0 Πi δL = 0.85) which reﬂects the gauge freedom. ∂ (7. ∂ (7. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. ˙ δ A0 δL = F i0 = E i .81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. as the vanishing of Π0 induces a constraint. 7.

Πν (y)] = 0. These problems are solved by the Lorentz constraint between physical states given above. pj ]P . i. λ). λ)c(k. 0)c(k.3) satisfy the conditions of a Lie algebra. and are equivalent with [c(k. Aν (y)] = [Πµ (x). i. (d) Use Poisson brackets to calculate q and p. c† (k ′ . Aν (y)]x0 =y0 = i gµν δ 3 (x − y).90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.e. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). It gives 3 k µ ǫ(λ) (k)c(k.e. which reproduce the Hamilton equations. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. j = 1. 0. Q]P .91) (7. q) that dO/dλ = [O. and [ℓi . where Q is the conserved quantity in Eq. The canonical equal time commutation relations are [Aµ (x). 2.94) Choosing k µ = (|k 3 |. 0. (b) Proof for a quantity O(p. pj ]P . for which it is suﬃcient that ∂µ Aµ |A = 0. a longitudinal photon and two transverse photons. µ λ=0 (7. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). λ)|A = 0. (7.93) which does exhibit problems with the time-like photon. (7. rj ]P . what are [ri . In fact the commutation relations imply ˙ [Aµ (x).92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. 7. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k. ˙ ˙ .11.89) with four independent vectors ǫµ .95) Exercises Exercise 7. 6. λ) − c† (k. containing a time-like photon.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. (7. ′ (7. λ)ei k·x . 0) . µ λ=0 (7.Quantization of ﬁelds 63 for physical states |A and |B . [ℓi . λ)e−i k·x + c† (k. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators.

. 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). (b) Check the above commutation relation [φ(x). It is invariant. (b) Show that ∆inhom (x) = 1 (2π)4 d4 k e−ikx 1 k2 − M 2 is a solution of the inhomogeneous Klein-Gordon equation. Pµ ] = i∂µ φ(x). (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. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. ∆hom (Λx) = ∆hom (x) if f is an invariant function.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. e−i a requires [φ(x). ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration). Give also the expression for ∆C . (2 + M 2 ) ∆hom (x) = 0. (e) Show that ∆R (x) = 0 if x0 < 0. we can write for ∆R an integral where the k 0 integration remains along the real axis. f (Λk) = f (k) for Lorentz transformations Λ.Quantization of ﬁelds 64 Exercise 7.3 (a) Show that the correct implimentation of symmetries in the Hilbert space.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.

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

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

−λ). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ) e−ik·x − d† (k. −λ)v(k. m) γ0 v(k. λ)γ0 v(k. op P (8. −λ). λ) e−ik·x − d† (k. iγ µ† ∂µ − M ψ(˜) = 0. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜).9) (8.7) = λ d k x x ηP b(k.10) i. the above operation reverses the sign of λ). x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0.11) (8. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ) e−ik·x − d† (k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). (i˜ µ ∂µ − M ) ψ(˜) = 0. λ)v(k. m). if helicity λ is used instead of the z-component of the spin m.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. .4) (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. m). x (8. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). x The latter behavior of the vector ﬁeld will be discussed further below.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). λ) P −1 = −η ∗ d(k. λ)u(k. (2π)3 2Ek (8. λ) e−ik·˜ + d† (k. m) = = ˜ u(k. λ)u(k. ˜ −v(k. 8.e. one can also consider the scalar ﬁeld and vector ﬁelds. 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. In the same way as the Fermion ﬁeld. λ)γ0 u(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. λ)v(k. γ x (8.5) (8.3) (8. λ) eik·x . λ) Pop = ηP b(k. −λ)u(k.6) (check this for the standard representation. ˜ Pop d(k.

21) where A is a 4 × 4 matrix acting in the spinor space. (8. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top .22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . antilinear operator is anti-unitary if A† = A−1 .e. Top |φ = φt | = (|φt )∗ . m). x iγ µT ∂µ − M ψ(−˜) = 0. λ). λ) e−ik·x + b† (k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.3 Time reversal xµ = (t. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k.18) (8. The (time-reversed) Dirac equation becomes. 8.15) (where one must be aware of the choice of spinors made in the expansion. λ)C v T (k. i. λ) Cop = ηC d(k.17) 3 = λ This shows that −1 Cop b(k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. m) ¯ C v T (k. As time reversal will transform ’bra’ into ’ket’. . m). x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. λ) e−ik·x + b† (k.Discrete symmetries 68 the latter being also a solution of the Dirac equation. Norm conservation requires Top to be anti-unitary2 .14) (8. we start with the complex conjugated x Dirac equation for ψ. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . (8. (2π)3 2Ek (8. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. it is antilinear1 .20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). λ). λ)u(k. r) −→ −˜µ ≡ −xµ = (−t. λ)v(k. λ)C uT (k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. r). λ) eik·x . to ﬁnd ψ t (−˜) that is a solution of the Dirac equation.16) d k ηC d(k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. x x (8. x (8. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. u(k. as discussed in section 4). writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). m) ¯ = = v(k. λ) −1 Cop = ∗ ηC b(k. x (8. The same relations hold for helicity states.19) Cop d(k.

−λ |¯.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. −p. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.32) (8. there are 16 independent of these bilinear combinations. λ) i γ5 C v(k. λ) Top = ηT b(k.2: The transformation properties of physical states for particles (a) and antiparticles (¯). op T (8. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k.Discrete symmetries 69 Table 8. a state |a.28) (8. (8. λ) e−ik·x (2π)3 2Ek (8. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ |¯. ˜ Top d(k. λ). λ). λ)u∗ (k. λ). Since there are 16 independent 4 × 4 matrices.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. −p. λ)iγ5 Cu(k. p. p. λ)v ∗ (k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.29) In table 2 the behavior of particle states under the various transformations has been summarized. −p. λ a |a.30) (8.24) (8. λ) eik·x + d† (k. λ)iγ5 Cv(k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). p. p. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. ∗ ˜ v (k.34) .31) (8. −p.25) (check this for the standard representation). λ)v ∗ (k. λ| ¯ C |¯. −λ a T a. 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. ˜ = 0|b(k)|A 8. λ) eik·x + d† (k.33) (8. λ)u∗ (k. λ)v ∗ (k.26) d k x x ηT b(k. λ a P |a. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) eik·˜ + d† (k. λ| a.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ) eik·x + d† (k. λ) T −1 = η ∗ d(k. x (8. λ)u∗ (k. λ). The result is = = λ (8.

and T. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).37) (8. s . (q 2 ). In many applications the expectation values of currents are needed. For instance for nucleons one has Γµ (p′ . This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .5 Form factors Currents play an important role in ﬁeld theory. in this case only q 2 = (p′ − p)2 . C. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).40) where the coeﬃcients are functions of invariants constructed out of the momenta. γ ν ]. (q 2 ). (8. s′ |V µ (x)|p. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x .Discrete symmetries 70 Table 8.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. p) can be built from any combination of Dirac matrices (1.38) (8. called form factors. 2M (8. ¯ In general the expectation value between momentum states can be more complicated. of which the expectation value between momentum eigenstates can be simply found.41) 2M 2M . F. γ µ ..g. as well as under the combined operation Θ = PCT (see Table 3). e. 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 ).39) where Γµ (p′ . σ µν .36) (8. (8. 8. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :.. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. As an example consider the vector current for a point fermion. ¯ ′ ′ (8. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). p′ |V µ (x)|p = u(p′ )Γµ (p′ . 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 . for the vector current V µ (x). C. The 16 combinations of Dirac matrices appearing above are linearly independent. F. p′ . p′ |V µ (0)|p = u(p′ )γ µ u(p). γ5 γ µ or γ5 ).35) The x-dependence can be accounted for straightforwardly using translation invariance. Note that the lagrangian density L (x) → L (−x) under Θ. p)u(p) e−i (p−p )·x .

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

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

in which the states are time-independent. (9.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. and the operators are time-dependent. in which the operators are time-independent. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .1 Time evolution as path integral U (t. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture.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. (9. AS (t) = AS and the states o are time dependent.3) (9. AH (t) = U −1 (t. H]. t′ . i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0.7) 73 . t0 ) = H U (t. [AH . QH (t)|q.4) ∂ U (t. t′ ≡ q ′ |q ′ . t0 )|ψS (t0 ) .Chapter 9 Path integrals and quantum mechanics 9. t0 ) AH (t0 ) U (t. t0 ).2) (ii) Heisenberg picture. t ′ QH (t′ )|q ′ . |ψS (t) = U (t. ≡ 0. t ≡ q|q.5) (9. t . Consider the two (time-independent) Heisenberg states: |q. (9. |ψH (t) = |ψH . (9. t0 ) ∂t (9. t ′ |q .

qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . H(P. tj+1 |qj . .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 . 2π (9.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . ˙ 2π 1 1 1 [A. t′ |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|.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . (9. The hamiltonian is an operator H = H(P. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . (9. Q) expressed in terms of the operators P and Q. tj+1 |qj ..q . B] + [A. which is a hamiltonian function in which the operators are replaced by real-numbered variables. 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 .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. B] + . t′ |q.16) this. B].8) Choose t as the starting point with |q = |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 . . Then qj+1 . QS |q ′ ≡ q ′ |q ′ .12) where the transformation between coordinate and momentum space involves q|p = ei p. 2π (9.14) .. ′ (9. [A. 2 12 12 (9. eA eB = eC with C = A+B+ (9. Combining the two terms gives qj+1 . n |q (9. t = q ′ |e−i H(t −t) |q .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. . use the Campbell-Baker-Hausdorﬀ formula. 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 . B]] + [[A. |q1 q1 |e−i H∆τ |q . . qj ). qj )]) .11) (9. with the correction1 being of order (∆τ )2 .

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

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

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. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). (9. θη = −ηθ.35) For applications to fermion ﬁelds. (9. F [θ. dx dy θ∗ (x) K(x.g. α(y) = δ(x − y). i. αy = δxy . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .32) (9.30) (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.33) (9. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. In principle the deﬁnitions of functionals is the same.e.31) δ δ α = dy α(y) = dy δ(x − y) = 1. y) θ(y) . It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. δα(x) (9. θ2 = 0. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . δα(x) δα(x) δ αKβ = dy K(x. which vanish. δα(x) or in discretized form δ 1 ∂ ∂ with . Examples of functional diﬀerentiation are (9. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . αn = δmn . the Gaussian-type functionals. however. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. products of the same pair appear.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ .29) (9. n θn fn (x) dx θ∗ (x) θ(x) . e. In the expansion of the exponential (from second to third line).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.34) (9. (9.38) .

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

Considering the source to be present between times t and t′ . T . n (9.s.e.65) J dq ′ dq Q′ . t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t|n n|Q. Consider. Before and after these processes there is only the vacuum or ground-state |0 . say.Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. t|Q. q) → L(q. t′ |q. t′ |q. t|Q.66) . t) φ∗ (Q) e+i En T . q . q) + J(τ )q] .5 The generating functional for time ordered products By introducing a source-term. harmonic oscillator. φn (q. t′ q ′ . which in turn are embedded between an early time T and a future time T ′ . t|n = q|n e−iEn t with q|n the time-independent wave function. T = n q. T ′ |Q. t = = ′ ′ V = q . L(q. ˙ ˙ physically pictured as. for g. i. t = ei Ht |q . q) + J(t) · q. a set |n of physical eigenstates. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. t |q. the creation of an electron (think of a radio-tube. t Q′ . T = n J t′ = = Dq exp i t dτ [L(q. The Heisenberg state |q. 9. q)] ˙ (9. T < t < t′ < T ′ . t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. furthermore. it is possible to switch on an interaction.62) p Dq D exp i 2π t dτ [pq − H(p.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. for the physical states o q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!).64) (9. t q. T with q. t is related to the Schr¨dinger state |q by |q. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. one has q ′ .e. q)] . ˙ J (9.63) for plane waves. T ′ |q ′ . making the electron) and the absorption of an electron (think of a detector). i.

t′ |q. . (9. Q(tn )|0 .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 ′ ). Better is the use ˜ of imaginary time t = −it. 9. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. thus T →i∞ lim q. t) (9. = (i)n 0|T Q(t1 ) . ˜ t′ → ∞.67) = = φ0 (q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. T ′ |Q.71) 9. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. t). T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .70) for the generating functional is in fact ill-deﬁned. t′ ) q ′ . T = φ0 (q. t|Q. J=0 (9. The importance of Z[J] is that the time ordered product of operators can be obtained from it. T → i ∞ and T ′ → −i ∞.68) (9. t 0 0out |0in J . J (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 ) . The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9.6 Euclidean formulation The above expression (Eq. . . dq dτ + J(τ )q(τ ) . eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . δJ(tn ) hence the name generating functional. . . Since we have (neglecting multiplicative factors).

i dq .2 Check the examples of functional derivation for real-.g.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. . det K Exercise 9. ZE [J] = = where LE q. ensuring convergence for the functional integral ZE [J].75) = − LE q.73) d˜ LE q.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors.and Grassmann-valued functions. As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources.n fm (x) Kmn fn (y).76) which is a positive deﬁnite quantity. . e. det K 1 .72) (9. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). + V (q).74) − V (q) 2 (9. ˜ dt (9. where Kp are the eigenvalues of the matrix Kmn in K(x. ˜ ˜ ZE [J] δJ(t1 ) . dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. δJ(tn ) J = 0 ˜ t = it (9. δ n Z[J] 1 Z[J] δJ(t1 ) . complex. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. τ dq ˜ dt ≡ −L q. for Grassmann valued functions. (9. (9. Exercises Exercise 9. 1 = √ . y) = ∗ m. . when L q. . In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . The diﬀerentiations in Z[J] and ZE [J] are related.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . .

z) = δ(x − z). det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .79) (9. y) K −1 (y. 2 (9.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .78) (9.3 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. .Path integrals and quantum mechanics 84 Exercise 9.

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

7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). These Green functions appear in the expansion Z[J] = n in n! d4 x1 .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. (2π)4 (k 0 + iǫ)2 − E 2 (10. . .8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). with Z0 [0] = 1.10) J=0 In order to determine ∆F . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. (10. . . . . . 2 2 → (10.13) . xn ) = x2 xn .e.9) i.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . (10. J(xn ). . . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. (10. d4 xn G(n) (x1 . 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). As discussed in the previous section (exercise 9. Furthermore.5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .

1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. (10. i. (2π)4 k 2 − M 2 (10. t) = 0 for t < 0.17) d4 k e−i k·x .20) = exp − 2 where or (see ﬁg. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. (10. . φ+ (0)] = (2π)3 2E are also shown in exercise (7. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation.18) Thus we see various solutions. (2π)4 (k 0 − iǫ)2 − E 2 (10. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . 10. ∆F = ∆C .16) C where C is a closed contour in the complex k 0 -plane.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x.15) e−i k·x d4 k . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). ∆(x) = − d4 k e−i k·x .14) They are solutions of the inhomogeneous equation. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.22) (10. µ 1 1 • Firstly. 4 k 2 − M 2 + iǫ (2π) (10. the advanced Green’s function ∆A (x) = satisfying ∆A (x. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).4). The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. t) = 0 for t > 0. (2π)4 k 2 − M 2 (10. 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 second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x).19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) .e.

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

Feynman diagrams for scattering amplitudes 89 10. (10.33) (4) i −i d4 z LI g 4! d4 z 1 δ i δJ(z) 1 exp − i 2 J ∆F J 2 −3 ∆2 (0) + 6i ∆F (0) F + d4 x ∆F (z − x) J(x) 4 d4 x ∆F (z − x) J(x) 1 exp − i 2 J ∆F J .26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10.31) G(2. x3 . Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. (10. x3 .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. x4 ) = G0 (x1 .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 . (10. (2) (0) (10.30) and in zeroth order G(0. x2 ) = G0 (x1 . ≡ −i g d4 z (10. the generating functional can be written. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .29) This expression will be the one from which Feynman rules will be derived. x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.g ) (x1 . x2 . with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].35) z . x2 . L (φ) = L0 (φ) + LI (φ).e. i.g 0 ) = G0 = 1.32) (10.g ) (x1 .34) Introducing a vertex point to which four propagators are connected. Z[J] = = Dφ exp i exp i d4 z LI (10. x2 ) = i ∆F (x1 − x2 ) = G(4.27) When interactions are present.

38) The result for the 4-points Green’s function is left as an exercise. . J=0 (10. . . which also was the only Green’s function for which the diagram was connected. (10. . d4 xn G(n) (x1 . G(n) (x1 . . . . i. In the free case we have seen that the exponent of Z[J] contained all essential information. xn ) J(x1 ) . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 .e. x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . . 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]. Diagrammatically this exponent only contained the two-point function. .41) in terms of socalled connected Green’s functions. .g ) (x1 .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. . J(xn ) c (10. This remains also true for the interacting case.37) G(2. .g 1 ) = 1 8 z 1 8 . (10. 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]. . did not contain parts that could be written as products of simpler Green’s functions. .40) in−1 n! d4 x1 .42) x1 G(n) (x1 . d4 xn G(n) (x1 . . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . . . . . (10. . . δJ(xn ) . . J(xn ). Z[J] = n in n! d4 x1 . . (10.39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. (10.36) J J from which one obtains G(0. . . xn ) J(x1 ) .

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

g. x4 ) = c 1 x3 z (10. For the free scalar theory. thus β. 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 . then (n) (n) Gsc = Gc for n ≤ 3. out| = eiϕ0 0. in = |p1 . in = β. the vacuum expectation value of the ﬁeld φ(x) is zero. considered explicitly. in into ﬁnal state particle states (out-states) |β. . . ∂µ ∂ µ + M 2 φin (x) = 0. in|SS † |α. out|α. implying the absence of tadpoles. Sβα ≡ β. out| and S † |α. in|S = β. φout and the interpolating ﬁeld φ(x). out .45) x2 as the only surviving diagram (see exercises). in ⇔ β. Proof: 0. pn . (10. out . in|S = 0. (10. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . in = |α. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). out|p′ . Next. x2 . p′ . . p. out = δαβ ↔ SS † = 1. in| (choose ϕ0 = 0). in|S|p′ . Proof: α. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout .e.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. out| ⇔ |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 ). out = |p′ . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y).46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. in = p. in = p. in = p. . in|S = α.47) . (2) The one-particle state is invariant (conservation of energy and momentum. (2) (10. e. out = p|p′ . out (suppressing except momenta all other 1 m quantum numbers). out|α. in = S|α. out|p′ . . x3 . x1 G(4.Feynman diagrams for scattering amplitudes 92 i. this will be translated to the action on ﬁelds. . in|p′ . translation invariance).44) J=0 For the interacting theory. x4 10.g ) (x1 . one ﬁnds for the connected 4-point Green’s function at order g. (10.

g. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. a) = S. in|φin S → φin S = S φout (x). out|φout = β.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. The relation between S and Z[J] To establish this relation. which in a later stage (renormalization) will become more important. e. a)SU −1 (Λ.e.51) (10. as it stands. √ Although the above.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.52) where Z is a constant. Proof: β. a causality condition that can be proven to imply the absence of interactions.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 ). (10. i. The convergence therefore must be weakened to t→−∞ √ (10.and out-ﬁelds is: φin (x) = S φout (x)S −1 . z) φ(z). (2π)3 2E (10. z) φ(z) d4 y d4 z ∆A (x − y) K(y. β. φ(y)] = Z [φin (x).62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . (10. The time evolution operator (see Eq. 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. The relation between S-matrix and in. the source term Jφ in the lagrangian density is treated in the interaction picture. φin (y)] = iZ ∆(x − y). S is invariant under e Poincar´ transformations: U (Λ. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . 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. 2 ∂µ ∂ µ + M0 φ(x) = J(x). (10.54) . implies the strong (operator) convergence φ(x) → Zφin (x). 9.53) 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). 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. this can actually not be used as it would imply [φ(x).49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions.

: =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. eB ] = [A. φ(x).e. [A. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. φ− f (10. t i Ut∞ [J] φ(x)U−∞ [J] (10. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). z) δU [J] . e = i. e e ] = [A. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. B]eB . just as we did for φ(x). (10. δU [J] δJ(x) = = √ i Z U [J] φin (x) − √ i Z φout (x) U [J] − d4 y d4 z ∆R (x − y) K(y. S U [J]] = √ Z δ S U [J].59) (10. φ(x). which is equal to the expression e in which the creation operators are placed left of annihilation operators. B] and [A. φ(y)] f (y) e e . provided [A.61) e φ+ f where the normal ordered expression : e φ f : is used. (10. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. z) . 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.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits.57) i [φin (x). δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. express it in terms of advanced and retarded Green’s functions.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). Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. B] is a c-number).63) . B + C]e e . z) d4 y d4 z ∆(x − y)K(y. y) δ δJ(y) : F [J]. we can. δJ(z) (10.62) where F [J] is some (arbitrary) functional.58) In order to ﬁnd a solution to this equation note that. C] are c-numbers. B] (for the case that [A. δJ(z) (10. one has [A.

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

= −i g . the factor 2 corresponds to the two-points Green’s function having two identical ends). For the symmetry factor consider the examples (given in G.27. Note that in calculating amputated Green’s functions external lines are neglected. to be precise iLI vertices in momentum space are introduced. ’t Hooft and M.27.Feynman diagrams for scattering amplitudes 96 (in fact the inverse of the operator found in the quadratic term is for real scalar ﬁelds also multiplied by a factor 2. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically diﬀerent connected diagrams. At these vertices each line can be assigned a momentum. which cancels the factor 1/2 in the quadratic piece. (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). Diagrammar) in the case of the interaction terms g f (10.66 and 10. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. but overall momentum conservation at a vertex is understood.69) LI (φ) = − φ3 − φ4 . For the interaction terms in the lagrangian. Veltman. or calculating full Green’s functions external lines are treated as propagators. Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). vertices and external particle wave functions in that diagram. for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar ﬁeld) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. If problems arise go back to the deﬁning expression for the generating function in 10.

S 3! 3! 2! 2 As a second example consider the diagram 1 2 There are three vertices. After connecting line 1 (6 ways) and line 2 (4 ways) we have 1 2 leaving 6 × 3 × 2 ways to connect the rest as to get the desired topology. Dividing by vertex factors and permutations of identical vertices. the result is 1 6×4×6×3×2 1 = = . i. there is no combinatorial factor like in the scalar case. (10.70) = exp i exp − In Feynman diagrams the propagator is still given by i∆F (k). denoted k = i k 2 − M 2 + iǫ Note that in this case the propagator does not have identical ends.e. . but it connects a source with its complex conjugate and therefore is oriented. External line 1 can be attached in six. The generating functional in the interacting case can be written as Z[J. Then there are two ways to connect the remaining lines such that the desired diagram results. 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 . which have been included in the deﬁnition of vertices (here 3! for each vertex). Finally divide by the number of permutation of the points that have identical vertices (here 2!). The total result is 6×3×2 1 1 = = . ∗ (z) i δJ(z) i δJ d4 x d4 y J ∗ (x) i∆F (x − y) J(y) .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. Divide by the permutational factors of the vertices. or equivalently as independent ﬁelds φ and φ∗ . after that line 2 in three ways. S 3! 3! 4! 2! 2 Complex scalar ﬁelds The case of complex scalar ﬁelds can be considered as two independent ﬁelds.

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

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 .Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. 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 . + 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.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). 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. writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . Vector ﬁelds and Quantum Electrodynamics As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). λ 1 ∂ (10. It is necessary to start at the end of a fermion line (with the above arrow convention an outgoing fermion wave function u or an incoming antifermion ¯ wave function v ) and keep on following that line. 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. 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 .

t2 (10. ¯ q2 (10.s4 e4 (m) µν (M) L L . 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 .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .76) 1 .s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s3 )(−ie)γ µ u(k. where u ¯ L(m) µν = = 1 2 u(k ′ . s′ ) ¯ u s.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.s2 . s2 ). To lowest order (∝ α = e2 /4π) only one diagram contributes.79) .5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ . s)γν u(k ′ . s)¯(k.s3 . s3 )γ µ u(k.s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10.Feynman diagrams for scattering amplitudes 100 10. s1 ) u e2 u(p′ . q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. s4 )γµ u(p. s2 )|2 ¯ q2 (10. s′ )γµ u(k. s1 ) ¯ −igµν u(p′ . s4 )(−ie)γ ν u(p.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.

the masses. x4 ) for the interacting case in φ4 -theory to ﬁrst order in the coupling constant g as obtained from the full expression for Z[J] in section 10.2. This is known as crossing symmetry.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s1 )γν v(k ′ . momenta and invariants are e− e+ → µ− µ+ scattering k p s t = = (k + k ′ )2 = (p + p′ )2 (k − p)2 = (k ′ − p′ )2 (k − p′ )2 = (k ′ − p)2 k’ p’ u = The scattering amplitude squared (spins summed and averaged) is given by |M |2 = 1 4s e4 s2 |¯(p.s4 = 1 2s × v (k ′ . . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. Similarly. Exercises Exercise 10. The diagram. s1 ) u(k. s2 )γµ u(k. for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. s4 ) u e2 v (k ′ .Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. s2 (10.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . s4 ) v (p′ . s4 )γ ν u(p. s3 )γ µ v(p′ . s2 )γµ u(k. s3 )γ µ v(p′ . s1 )|2 ¯ s 1 . (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive.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 . s3 ) ¯ ¯ 1 . (4) (4) . To lowest order (∝ α = e2 /4π) only one diagram contributes. (c) Do the same for the connected Green function Gc . . s2 ) ¯ ¯ 1 2s u(p. .s2 .s2 3 .s3 .

. It is of the form −i M = A1 − A2 . + pn ) G(n) (p1 .2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a.Feynman diagrams for scattering amplitudes 102 Exercise 10. (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . . . . xn ) ∝ (2π)4 δ 4 (p1 + . . . . . . + pn ).4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. xn + a) = G(n) (x1 . . hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . Exercise 10. in the amplitude (Tij = A∗ Aj ). |M |2 = T11 + T22 − T12 − T21 . Pictorially this implies. . . Show that i the amplitude is symmetric under the interchange of t ↔ u.38. e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . . . pn ). . . . for instance. or if we in general would consider all momenta as incoming. . . t and u. xn ). 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).3 Show that the combinatorial factors found using the rules given in section 10. 1 2 (b) Calculate the quadratic pieces and interference terms. but the incoming and outgoing momenta are identical. which means overall momentum conservation in Green functions in momentum space. . . . . . Exercise 10.37 and 10.3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. . . their sum is zero. pn ) only has n − 1 independent variables in it. Express the contributions in invariants s. . xn ) ≡ (2π)4 δ 4 (p1 + . Note that the function G(n) (p1 . .

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

trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 2mb (11. 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. p1 . m2 . m2 . = 2 s λ(s. a (11. m2 . m2 . 4π s 4π 8π s 4π (11. p2 ) = = where λ12 denotes λ(s. 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 √ = . which in the speciﬁc case a b also can be expressed as λ(s.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . 2mb λ(s.10) The function λ(s. ptrf ). 4s (11.12) pb = (mb . m2 ) is a function symmetric in its three arguments. mb and s).8) (11. m2 ) = 4(pa · pb )2 − 4p2 p2 . for instance. 0). 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.16) .14) One can. m2 ). (11. m2 ) a b .9) (11. m2 . 2 s s − m2 + m2 a b √ .11) (11.13) (11. p1 . Also in this case one can express the energy and momentum in the invariants.15) Explicit calculation of the reduced two-body phase space element gives dR(s. In that case trf pa = (Ea . 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 ) a b .

26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. The initial state. The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . An often used set of invariants are the Mandelstam variables..4). E . = (11. The various possibilities are referred to as diﬀerent reaction channels. 2s (11. pT = (px .. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.21) (11.17) (11.27) with m2 = m2 − p2 = m2 + p2 + p2 . usually is a 2-particle state. . for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. Examples appear in → π− + p (11. py ). H1 + H2 → h + X. a c with q = λab /4s and q ′ = being 0 or 180 degrees.22) (11. The variable s is always larger than a c b d the minimal value (ma + mb )2 . Of course there are not only 2-particle channels. 11. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .19) (11. in which one particle is detected. tmax min λcd /4s. 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 . pc . in exclusive measurements all particles are detected. (11. the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . Consider the 1-particle inclusive case. however. Consider the process a + b → c + d.18) (11. there is usually a preferred direction. At high energies.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. pz = mT sinh y. In inclusive measurements no particles are detected in the ﬁnal state. pd ) = 4π s 4π 8π s 4π dt dt √ √ . dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. usually many particles are produced.25) dR(s.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. A speciﬁc reaction channel starts contributing at the threshold value (Eq. Instead of the scattering angle.20) → π0 + n → π+ + π− + n → . Writing E = mT cosh y. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels.

u = (p1 − p4 )2 = (p2 − p3 )2 . For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. t. c bd Mad→c¯(su = u.2 Crossing symmetry In the previous chapter. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. uu = s) = Mab→cd (s. ¯ 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.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. t. ut = u) = Mab→cd (s. 11. u). t. (11. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. t = (p1 − p3 )2 = (p2 − p4 )2 . u) with s = (p1 + p2 )2 = (p3 + p4 )2 . u). With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η.31) . (11. η = − ln(tan(θ/2)) = tanh−1 (cos θ).Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. 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. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). which means that rapidity distributions dN/dy maintain their shape. uu = (pa − p¯)2 = s. tt = s. ¯ tu = (pa − pc )2 = t. ¯ b (11. tu = s. ¯ tt = (pa − p¯)2 = s. b ut = (pa − pd )2 = u. t-channel and u-channel processes respectively. This is known as crossing symmetry.

say b) the cross section σ(a + b → c + . The proportionality factor has the dimension of area and is called the cross section.3 Cross sections and lifetimes Scattering process For a scattering process a + b → c + . t.e. To see this one can make a two-dimensional plot for the variables s. (consider for convenience the rest frame for the target. .) is deﬁned as the proportionality factor in Nc = σ(a + b → c + . 1 N . T where V and T indicate the volume and the time in which the experiment is performed. . 2 2 2 2 u(t − s) + (Ma − Md )(Mc − Mb ) . t and u.Scattering theory 107 One can also phrase it in the following way: one has a single analytic function Mab→cd (s. Nb indicates the number of (target) particles b. . This deﬁnes the boundaries of the physical regions. . (11. u) that represents physical amplitudes in the physical regions for three scattering processes. Nc /T indicates the number of particles c detected in the scattering process. 2 2 2 2 t(s − u) + (Ma − Mc )(Mb − Md ) .32) σ= trf T · V ρa ρb va co . i. while the ﬂux of the beam particles a is trf ﬂux(a) = ρa · va . This is a consequence of the constraint 2 2 2 2 s + t + u = Ma + Mb + Mc + Md . . shown below for the case of equal masses. which for a density ρb is given by Nb = ρb · V . . 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 ) . _ ac cos s= 0 _ bd cos θ t =1 θt =− 1 t u cos θ u = 1 sθ u=− s t=0 cos θ s = 1 u= 0 _ ad _ cb ab 1 cd 1 cos θs =− 11.) · Nb · ﬂux(a).

39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). as expected. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. .e. (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. dt (11. (2π)4 δ 4 (Pi − Pf ) 2 = (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf ) d4 x ei(Pi −Pf )·x V. . pn ). 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 . .T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). σ= √ 2 λab T · V (11. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .38) (in which we can calculate −iMf i using Feynman diagrams). .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. For the (proper) decay width one thus has 1 N Γ0 = . Microscopically one has Ndecay = Na · Γ T or Γ= N 1 .35) This quantity is not covariant. The decay time for moving particles τ is related to the decay time in the rest frame of that particle (the proper decay time τ0 ) by τ = γ τ0 . N 1 . N and T · V are covariant.34) i. it is. (11.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. T · V ρa (11. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. .Scattering theory 108 Although this at ﬁrst sight does not look covariant.36) (11.T V.

(11.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. 11.43) dΩcm = π M (s. . p1 . note the sign and cos θ = qi · qf ). t) λab 4π dt. . ˆ ˆ Inserted in the unitarity condition for M .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 . (11. θ) in quantum mechanics. θ) = −8π s ℓ (in analogy with the expansion for f (E. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). θcm ) √ q 8π s 2 2 (11. . .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). (11. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E.e. pn ). implies for the scattering matrix M . . p1 .46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. respectively. 16π 2 s (11. θ).41) (11.47) M (s. 2 λab and for instance for two particles dσ = q ′ M (s. for which one has dσ/dΩ = |f (E. (11. i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . θ)|2 .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. q n ). (11.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. i. The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. q f ). The result is N = |Mf i |2 dR(s. . . 8π s 2π 8π s . p1 . θ) or M (q i .42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. . . . pn ). pn ) |M |2 q 32π 2 m2 dΩ |M |2 . for which the √ amplitudes squared have been calculated in the previous chapter. . δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . q n ) q n qn √ Mni (q i . see 11.

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 . n (11. (11. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ).54) ℓ .52) From the imaginary part of M (s. (11. 2i q 2i q (11. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = .50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. (11. 0).Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i .53) The diﬀerence is the inelastic cross section. ℓ (11. Using σ= dΩ q ′ M (s. q ˆ i. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . θcm ) √ q 8π s 2 .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 . the total cross section can be determined.e. (11.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .49) or Im Mℓ = q |Mℓ |2 . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). In general a given channel has |Sℓ (s)| ≤ 1. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ).

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

This is a narrow resonance discovered in 1974. 2 q (s − M 2 )2 + M 2 Γ2 (11. Note that because of the presence of a background the phase shift at resonance may actually be shifted.Scattering theory 112 (Prove this using the unitarity condition for partial waves). u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . t)} . Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. u) = g(t.2 GeV. M2 − s (11. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . All these decays can be seen and leave an ’invisible’ width of 498 MeV. the partial width into e+ e− is Γee = 5. From this one sees that a resonance has the same shape in all channels but diﬀerent strength.26 keV. the full width is Γ = 88 keV. Exercises Exercise 11.59) reaching the unitarity limit for s = M 2 .49 GeV. u) + f (u. The half-width of the resonance is M Γ. Knowing that each neutrino contributes about 160 MeV (see next chapter). The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ .e. The cross section at resonance is about 30 nb. • The J/ψ resonance in e+ e− scattering. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. dt s(s − 4m2 ) with f (t. where furthermore σinel = 0. Three neutrinos explain the resonance shape. t) + g(u.88 MeV. Its mass is M = 3096. This characteristic shape of a resonance is called the Breit-Wigner shape. which is attributed to neutrinos. At resonance the cross section σT (π + p) is about 210 mb.4) can be written as 4πα2 dσ = {f (t.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. a fast change in the phase shift for a narrow resonance. • The Z 0 resonance in e+ e− scattering with M = 91. 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). u) + g(t. its width Γ = 120 MeV.1 Show that the cross section for electron-electron scattering (exercise 10. Limiting ourselves to a resonance in one channel. Γ = 2. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. i. Its mass is M = 1232 MeV.

6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0.3 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.2 Calculate the ﬂux factor. writing N (s) M1 (s) = 2 + iM Γ .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∆ .000 123 Γ.999 877 Γ and Γ(π − → e− νe ) = 0. Exercise 11. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . a quantity which we will ¯ ¯ encounter in the next chapter (section 12. As discussed for unstable particles.8. Thus. cm |va − vb | in the center of mass system for two colliding particles a and b. Compare the results with the experimental π − lifetime τ = 2. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. ρcm ρcm a b 1 cm . Exercise 11.2).Scattering theory 113 Exercise 11. Calculate now fπ . we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. 4.

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

which connects two identical vectors. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. should be independent of such transformations.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). However. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).16. (12. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). = . but expresses them with respect to diﬀerent bases. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x).The standard model the basis for x G and x+dx 117 G. In principle such a phase in a given point is not observable. i.19) This gives rise to a (path dependent) phase in each point. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . 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. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). (12.e. -dx -dy µ µ dyµ µ dx For a vector.14) (12. ds which is solved by dψ(s) ds = dxµ . (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). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. (12.15) ψ(x). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).e. i.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. 12. The relation in Eq.

φ= 2 φ3 (12. φ1 φ .20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . i. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. Dν ]. Nevertheless. a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components. 12. 2 2 4 −V (φ) The potential V (φ) is shown in ﬁg. In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν . This characterizes spontaneous symmetry breaking. the groundstate itself appears degenerate. i>j which is rotationally invariant.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0.2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. e. and equivalently Aµ can be considered as a pure gauge eﬀect. As there can be one and only one groundstate. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect). this means that there is more than one possibility for the groundstate. Similarly as the deﬁnition of the covariant derivative the eﬀect is thus frame-independent and we have the transformation law Gµν → S(x)Gµν S −1 (x).The standard model 118 where Gµν = (i/g)[Dµ . usually being (slightly) prejudiced by impurities.e.2. If it is zero one has dxµ Aµ (x) = 0. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . in reality a not perfectly symmetric situation. Only if this quantity is nonzero a physically observable eﬀect of Aµ exists. If the temperature is below a certain critical temperature (T < Tc ) the kinetic energy is no longer dominant and the above hamiltonian prefers a lowest energy conﬁguration in which all spins are parallel. .g. external magnetic ﬁelds. Nature will choose one. 12. we can disregard those ’perturbations’ and look at the ideal situation. In this case there are many possible groundstates (determined by a ﬁxed direction in space).

21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero).25) 2 2 2 Therefore there are 2 massless scalar particles. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld.22) F Realization of symmetries (model independent) In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented. . In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . . etc.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . They are known as the Weyl mode or the Goldstone mode: Weyl mode. i. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. 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. (12.e. The situation now is the following. say 0 0 . The η massless particles are called Goldstone bosons. Any constant ﬁeld with ’length’ F is a possible groundstate. corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . therefore a choice must be made. ϕc = 0|φ |0 = (12. (12. for the vacuum Qa |0 = 0. while the lagrangian including the nonzero groundstate expectation value chosen by nature. while the quadratic terms must correspond with non-negative masses. .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. In the case of degeneracy. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . In this situation one speaks of spontaneous symmetry breaking.) terms constitute interaction terms. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. 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. (12. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . ϕ2 and η. has less symmetry. In this case the spectrum is described . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. The classical groundstate appears degenerate. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . A quantum ﬁeld theory has only one nondegenerate groundstate |0 .23) F +η the latter only forming a minimum for m2 < 0.

= ψ(i/ − M )ψ. Taking the derivative. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). T = τ /2. Neglecting at this point the local color symmetry.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. .30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. Goldstone mode.e. (12. Using the Dirac equation. . a ∂ µ Jµ = 0. i. ψu mu md ψ = ψd ..32) (12. Including a sum over the diﬀerent ﬂavors (up. . The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).31) . .e. and one must have mπa = 0. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. M = . strange. V µ = ψγ µ T ψ. i. 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) (12.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). i. The generators Qa (in that case the rotation operators Lz . etc. (12. it is easy to see that one gets ∂µ V µ = i ψ [M. ψ −→ ei α·T ψ. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. [Qa . a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . called ﬂavors. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. irrespective of the fact if they annihilate the vacuum.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). denoted |π a (k) . π (12.26) for all the states labeled by a corresponding to ’broken’ generators. Lx and Ly or instead of the latter two L+ and L− ) are used to label the multiplet members or transform them into one another. a and a′ are degenerate states. down.) one can write L = f ψ f (i/ − Mf )ψf . T ] ψ. (12. a massless Goldstone boson for each ’broken’ generator. H] = 0. there must be a nonzero expectation value 0|Jµ (x)|π (k) . As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. A bit more formal. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . This means that they are operators that create states from the vacuum. ∂ ∂ (12. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .The standard model 120 by degenerate representations of the symmetry group. if the generators Qa generate a symmetry.e.

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

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. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . After symmetry breaking the same number (as expected) comes out. .e.37) (12. made into a gauge theory. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. (12. The symmetry is broken in the same way as before and the same choice for the vacuum. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. F which gives for the lagrangian density up to quadratic terms L 1 1 λ 1 = − Gµν · Gµν + Dµ φ · Dµ φ − m2 φ · φ − (φ · φ)2 4 2 2 4 1 2 2 1 1 2 = − (∂µ Wν − ∂ν Wµ ) · (∂ µ W ν − ∂ ν W µ ) − g F Wµ W µ 1 + Wµ W µ 2 4 2 1 + (∂µ η)2 + m2 η 2 + . like a photon. Note that the number of massless gauge bosons (in this case one) coincides with the number of generators corresponding to the remaining symmetry (in this case rotations around the 3-axis). (12.41) 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 . 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. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). again 9 degrees of freedom. in this case three-dimensional) representation reads (La )ij = −i ǫaij . . . (12.40) a Dµ φ = ∂µ φ − ig Wµ La φ. 0 0 0 = . is made. 0 ϕc = 0|φ |0 = 0 . two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . while the number of massive gauge bosons coincides with the number of ’broken’ generators.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). thus 9 independent degrees of freedom.The standard model 122 12. but one has 1 massless gauge ﬁeld (2). having two independent spin components) and three scalar ﬁelds (one degree of freedom each). 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. i. Initially there are 3 massless gauge ﬁelds (each.39) (12. . One may wonder about the degrees of freedom. as in this case there are no massless Goldstone bosons. We have the possibility to perform local gauge transformations. 4 2 where Since the explicit (adjoint. 0 φ= (12.

The standard model 123 12. the leptons (elektrons. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. etc.50) (12. ∂ ∂ ∂ (12.and righthanded particles. however. explicitly L −→ e−i β/2 L. R −→ e−i β R. 2 2 = ∂µ R + i g ′ Bµ R. = ∂µ L − (12.44) which has an SU (2)W symmetry under transformations eiα·TW . τ ). 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. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. i. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. muons. 2 (12. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. ψR/L = 2 (1 ± γ5 )ψ decouple. electromagnetic and weak forces. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . R. neutrinos. This is written as 3 Q = TW + YW . under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 .e. (12. the quarks (up.47) and YW is considered as an operator that generates a U (1)Y symmetry.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. The procedure.) and the gauge bosons responsible for the strong.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. νµ and ντ ) and their antiparticles. 1 As they are massless.). down. U(1)Y U(1)Y (12. muon µ− or tau τ − ).45) (12. its corresponding neutrino (νe . (12. while the lefthanded particles form a doublet. that gives a good description of the physical world.51) .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 /2.48) (12. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. left. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. etc.

59) . Putting this in leads to L (f ) =L (f 1) +L (f 2) . For the classical ﬁeld the choice θ4 = v is made. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. † i i g Wµ · τ − g ′ Bµ )φ. −V (φ) L and (h2) = −Ge (LφR + Rφ L). leading to the parametrization 1 0 φ(x) = √ (12. . with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. TW = ±1 or 0 and YW = 0 (thus Q = TW ) and 3 Bµ is a singlet under SU (2)W (TW = TW = 0) and also has YW = 0. (12.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. this leads to the lagrangian L (h1) . . the U (1)Q symmetry (generated by the charge operator). (12. h) (12. Using local gauge invariance θi for i = 1.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .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. 4 4 In order to break the symmetry to the symmetry of the physical world.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. . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). . (12.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . L where L (h1) (h) =L (h1) +L (h2) .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 . + 8 (12.58) (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. .

65) The weak mixing angle is related to the ratio of coupling constants.66) From the coupling to the photon. 4 2 MW g2 v2 = . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . .67) The coupling of electrons or muons to their respective neutrinos.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 . 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 + . 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 . (12.64) (12. g ′ /g = tan θW . 2 Zµ (12.The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.61) (12.69) . 2θ 2θ 4 cos W cos W 0.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . and correspond to three massive particle ﬁelds (W ± and Z 0 ) and one massless ﬁeld (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 . 2 (12. . we can read oﬀ e = g sin θW = g ′ cos θW . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. (12.63) (12.60) (12.

leptons with CV = −1/2 + 2 sin2 θW ≈ −0. 3 = TW .g.231 2. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).40 GeV. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.166 4 × 10 0. me /v is extremely small. (12.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.80) First. 91. Finally we want to say something about the weak properties of the quarks. The coupling to the Higgs particle is weak as the value for v calculated e.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. 1. i. (12. . 2 2 2 (12. 80. g ′ and v determine a number of experimentally measurable quantities. one has the relation g2 e2 1 GF √ = . Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). muon or tau. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . giving rise to a running coupling constant after renormalization of the ﬁeld theory.e. v (12.5 MeV (exp.71) GeV −2 = = = = .19 GeV.72) (12. from the MW mass is about 250 GeV. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.74) (12. Comparing this with the total width into (invisible!) channels.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 . 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. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. g2 MZ 2 2 (CV + CA ). (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. i.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. (12.73) (12. −5 (12.76) with CV CA 3 = TW − 2 sin2 θW Q. With the choosen vacuum expectation value for the Higgs ﬁeld.e. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.77) (12.

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

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 .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. † Λu Λ† = Vu G2 Vu .227. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .g. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. i. . UR = uR cR tR .83) where Vu and Vd are unitary matrices. e.22 and η ≈ 0. (12. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. the weak mass eigenstates are L (h2. respectively). A ≈ 0.34. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .e. One then reads oﬀ that starting with the family basis as deﬁned via the left doublets that the mass eigenstates (and states coupling to the Higgs ﬁeld) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . ms and mb ). (12. u u where we include now the three lefthanded quark doublets in QL . d d (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 12.82) space. each of these containing the three families. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . The Higgs ﬁeld is still limited to one complex doublet.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices.86) the unitary CKM-matrix introduced above in an ad hoc way. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0.82 and ρ ≈ 0. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .

The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. The mass ﬁelds ER † mass † mass = We ER . and obtain L (h2.92) . we ﬁnd massless neutrinos.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. EL = Ve EL . (12. one can write Λe = Ve Ge We and we ﬁnd † L (h2.5×10−3 eV2 .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .93) 1 = 0 0 0 c23 −s23 0 s23 c23 c13 0 −s13 eiδ 0 s13 eiδ 1 0 0 c13 c12 −s 12 0 s12 c12 0 0 0 1 .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .89) with three righthanded neutrinos added to the previous case. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2.e. i. UPMNS a parametrization that in principle also could have been used for quarks. = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). decoupling from all known interactions. The mixing is intriguingly close to the Harrison-Perkins-Scott tri-bimaximal 23 mixing matrix 0 0 2/3 1/3 0 2/3 1/3 0 1 1/2 − 1/2 − 1/3 UHPS = 0 1/3 − 1/2 . (12. For neutrino’s this matrix is parametrized in terms of three angles θij with cij = cos θij and sij = sin θij and one angle δ.ℓ) = −LφΛe ER − ER Λ† φ† L. In this case. e N L= L EL (12.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. As before. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . (12. mµ and mτ . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . there is no family mixing for massless neutrinos. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . EL = Ve† EL . (12. since the weak current is the only place where they show up.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. 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 .

For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. hence the mixing matrix becomes iα /2 0 0 e 1 (12.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. The mass matrix can be written as ML |MD | eiφ M = (12. For the leptons. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. Actually.94) M L NL NL + M L NL NL . it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ. the massless and massive Majorana neutrinos. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).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.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . the left. a unitary matrix relating them to the weak eigenstates.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. 1 ∗ c c (12. one being massive. are equivalent to two decoupled Majorana neutrinos (see below). (12. 1 L Υ ′ = nR + nc . − MD MR nR 2 (12. 0 0 1 L mass.c.99) . Thus 2 Υ ′ = nc + nL . however. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. We use here the primes starting with the weak eigenstates. Thus (disregarding family structure) one has two Majorana neutrinos.ν =− containing three (CP-violating) phases (α1 . This is called the see-saw mechanism (outlined below). For these light Majorana neutrinos one has. 2 R ψ = nR + nL . 1 c ∗ c L mass.98) (12.and righthanded species then just form a Dirac fermion.95) 2 In order to have more than a completely decoupled sector. α2 and δ). with for the righthanded sector a mass term MR . The way to circumvent this is to introduce as in the previous section righthanded neutrinos. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. (12. 2 although this option is not attractive as it violates the electroweak symmetry. Absorption of phases in the states is not possible for Majorana neutrinos. For the neutrino sector. 12. coupled by a Dirac mass term. is to add in Eq. as above. couple the right. one must for the neutrinos as well as charged leptons.ν = − MR NR NR + MR NR NR .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.

For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . T ] ψ. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . Prove that if the generators Qa generate a symmetry. 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 .e. 2 with real masses Mi . i ψ {M. Exercise 12. H] = 0.2 Derive for the vector and axial vector currents. i.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 . Exercise 12. (12.102) = 1 2 2 ML + MR + 2|MD |2 2 ± 2 2 2 2 (ML − MR )2 + 4|MD |2 (ML + MR + 2ML MR cos(2φ)) .1 Consider the case of the Weyl mode for symmetries. a and a′ are degenerate states. L Υ2L nR Υ1 and Υ2 . Exercises Exercise 12. related via c Υ1R = U nL .The standard model 131 taking ML and MR real and non-negative. i.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . T } γ5 ψ. (12. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. .101) Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2 . 2 U (M M † ) U † = M0 .e. 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). [Qa . and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .105) for i = 1. (12. Υ2R nR (12.

) are proportional to the mass squared for bosons or mass for fermions of the particles. Γ(Z 0 → νe νe ).4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 .6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . at least up to a high (which?) order in λ. . (b) Calculate the width to electron-positron pair.7 In this exercise two limits are investigated for the two-Majorana case. Exercise 12. Γ(Z 0 → e+ e− ). Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . 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. Exercise 12. e+ e− h.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . 2 2 The masses are mt ≈ 175 GeV. and the width to a pair of neutrino’s. etc. to be precise one has for bosons (b) or fermions (f). the weak mixing angle is given by ¯ sin2 θW = 0. mb ≈ 5 GeV and MW ≈ 80 GeV. ZZh. 3 TW . The mass of the Z 0 is MZ = 91 GeV. Mf M2 and ghf f = . ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld.5 Show that the couplings to the Higgs (for W + W − h. Exercise 12.231. Exercise 12. hhh.

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

relativistic quantum mechanics

relativistic quantum mechanics

Are you sure?

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

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd