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

# QUANTUM FIELD THEORY

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

November 2011 (version 6.04)

Contents

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

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

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

1

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

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

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

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

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

Zuber. Quantum Field Theory. Peskin & Schroeder and Srednicki.1 5. These notes Section 1. M.3. Itzykson and J. The notes contain all essential information. Quantum Field Theory. 3.2 36 22 .3 3.5 5.3 3.1 4. 2.1. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. S.2 3. Cambridge University Press.-B.7 2.1 3.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. C. Diagrammatica. Ryder. Cambridge University Press.2 3. Vol.4 6.4 4. Veltman. The quantum theory of ﬁelds. 4.2 2. Weinberg.3 3.V.3 3.3 6.2 3.1 2 2 3.3 2. Addison-Wesly. Cambridge University Press.2 4. 1995. I: Foundations. 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). 2007.4 3.4 3. 6.3 References As most directly related books to these notes.1 6.3 2. 3. Srednicki. Quantum Field Theory.E.1 1. I refer to the book of Srednicki [1] and Ryder [2]. L.3 2. Corresponding chapters in books of Ryder.2 6.4 3. Cambridge University Press. 5.H. but are rather compact. 1995.8 3. 1996. Schroeder. M.2 2. An introduction to Quantum Field Theory. Vol. M. 3.2.2 3.1 2. 1994.2 3.2 6.3 2. 1980.2 2. Peskin and D. McGraw-Hill.1 2. 2.2 1. 1. II: Modern Applications.3 4. Cambridge University Press. 1985.5 2.3 3.4 2.5 4.4 2.3 2.5 2.1 3.

2 11.4.3 7.5 7. 6.7. 6.6 9.5.3. 4.1 7.8 5 4.5 4.3 6 8 9 Srednicki 3 3 3 37 5.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.2 8.4 3. 6. 6.3 3. 2.2 12.2.4 8.3 11.5 11. 6.3 9.3.5 4.7 5.4. 4.2 4.1 9.2 7.6 3.4 9.1.8 6.6 3.3 2.4 12.6 6.6 8.6 3. 4.3 10.5 .2.2 5.3 20.2.1 8.4 10.2 9.1 9.1 20.3.4.4 4. 9. 4.3 6.1 12.5 2.1 20.3 8.6 10.1 20.5 9.5 9.7 4.5 7.2 10.3 12.4 These notes Section 7. 8.2. 9. 20.5 6.5 6 7 6 6.1.4 2.2 8.1 5.1 10.1.3 7.6 4.4 7.4 12. 2.3 8.1 11.10 8.

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

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

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.

One can distinguish: • ds2 > 0 (timelike intervals). independent of a coordinate system.22) (1. ˆ ˆ (1. The real.νn can be conν structed. −x). Because of the diﬀerent signs occurring in gµν . The length (squared) of a vector is obtained from the scalar ˆ product. 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.Introduction 5 that can be used in the cross product of two vectors z = x × y. . and are given by (x0 . δim δjm δkm δin δjn δkn . it is convenient to distinguish lower indices from upper indices.. the invariant length or distance (squared) is not positive deﬁnite.26) (1. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . x2 . such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . xi are the three space components.21) (1. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . x2 = x · x = xµ xν nµ · nν = xµ xν gµν . one must distinguish tensors with upper or lower indices and one can have mixed tensors. (1. When 3-dimensional space is considered as part of Minkowski space. however. are . In special relativity we start with a four-dimensional real vector space E(1. x3 ) = (t. the points lie on the lightcone and they can be connected by a light signal. linear transformations that do not change the length of a four-vector are called the Lorentz transformations.2. . x2 . • ds2 = 0 (lightlike or null intervals). In Minkowski space. x) = (x0 . x1 .3). given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). x1 .. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations.24) The quantity gµν ≡ nµ · nν is the metric tensor. The (invariant) lengths often have special names. where the coordinate t = x0 is referred to as the time component. denoted as V ′µ = Λµν V ν .. (1. Relations relating tensor expressions. xµ = gµν xν . in this case an inertial system exists in which the two points are at the same time and ds2 just represents minus the spatial distance squred ds2 = −dx2 . These transformations do change the components of a vector. Λµn n T ν1 . Vectors are denoted x = xµ nµ .3) with basis nµ (µ = ˆ 0. Many other four vectors and tensors transforming like T ′ µ1 . in which case zi = ǫijk xj yk .µn = Λµ1ν1 .23) δjm δkn − δjn δkm .1. in this case an inertial system exists in which the two points are at the same space point and in that frame ds2 just represents the time diﬀerence ds2 = dt2 . The Poincar´ transformations include e e Lorentz transformations and translations. We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. • ds2 < 0 (spacelike intervals).25) Within Minkowski space the real. we will use upper indices for the three-vectors.. So we could have used all upper indices in the above equations. x3 ). The lower indices are constructed in the following way. No diﬀerence is made between upper or lower. One has x2 = xµ xµ = t2 − x2 . Unlike in Euclidean space.

The length squared of ∂ is the d’Alembertian operator. (1. ∂2 . g0 = g1 = g2 = g3 = 1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. constructed via the metric tensor itself. . Each of these permutations leads to a minus sign. . (1.Introduction 6 called covariant. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). starting from ǫ0123 = 1. explicitly written as (p0 . but more important has the same eﬀect on lefthandside and righthandside.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.28) ν ν Note that gµ with one upper and lower index. It is an invariant tensor.33) (1. = −24. 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. (1. ∂3 ) = ∂ ∂ ∂ ∂ .34) (1. = ∂ . a∞ = 4πǫ0 2 /me e2 . .27) It is easy to convince oneself of the nature of the indices in the above equation. p) = (E. where m is called the mass of the system. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . for instance.∇ . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . because one has (1. not aﬀected by Lorentz transformations.32) . Note that in this equation one has on left. p).36) Exercises Exercise 1. (1. deﬁned by ∂2 − ∇2 . ∂1 . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. (1. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . ∂t (1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . ′ ′ . 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.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. 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 .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ .

n1 . (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). 2. 0.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ .511 MeV. n1 . . a+ . 0. One does not need to know . Also for the coordinates [a− .71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). 0. ˆ ˆ .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . c. 0. 1. a1 . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . by a simple few-line reasoning [For instance: If {µ. g+− and g−+ .2 Prove the identity A × (B × C) = (A · C) B . σ} is a permutation of {0. a2 ] we can ﬁnd basis vectors n− . n2 . ˆ ˆ ˆ ˆ These are n1 = (0. a2 ] or [a− . (d) Use the value of the gravitational constant GN / c5 = 6. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . Also construct and calculate the Planck length Lpl . Exercise 1. Compare it with the proton mass and use Eq. 0) ˆ n2 = (0.13 to give its actual value in kg. a2 . 0. This demonstrates how a careful use of units can save a lot of work. a+ . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . (b) The coordinates (a0 . n+ . a1 . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. [Note: what is the MKS unit for V/T?] Exercise 1. How are a+ and a− related to a+ and a− . ]. 0) ˆ n0 = (1. 3} the index α can only be equal to one of the indices in ǫµνρσ . n3 . 1. 0. me . a1 . 0) ˆ n3 = (0. but only appropriate combinations.3.(A · B) C using the properties of the tensor ǫijk given in section 1. ν. α and me c2 = 0. ǫ0 . n2 . Exercise 1. re = e2 /4πǫ0 me c2 to the Compton wavelength. 1. 1) ˆ . ρ. . g−− .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). a+ . e. 1. a3 ) are the expansion coeﬃcients using the basis vectors n0 . which is the Compton wavelength for the Planck mass.

8 . i ∂ ∇2 ψ(r. φk (r. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. written as pµ −→ i∂µ is covariant (the same in every frame of reference). t) = exp(−i k 0 t + i k · r). t) = − ψ(r. 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. 2 + M 2 φ(r.3 are not covariant.2. namely plane waves characterized by a wave number k. t) = ∂2 − ∇2 + M 2 φ(r. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.3) Equations 2. (2. But the replacement 2. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. ∂t 2M (2. ∂t2 (2.Chapter 2 Relativistic wave equations 2. p). Although it is straightforward to ﬁnd the solutions of this equation. p2 = pµ pµ = E 2 − p2 = M 2 . ∂t p −→ pop = −i∇.2) Acting on the wave function one ﬁnds for a free particle.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . t). E = p2 .6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ.1 and 2.5) with (k 0 )2 = k2 + M 2 . (2. one obtains (2. 2M (2. t) = 0.4) where M is the particle mass.1 The Klein-Gordon equation In this chapter. (2.7) k 0 = ± k2 + M 2 = ±Ek .

for which we need the interpretation of φ itself as an operator. ∂ρ = −∇ · j. The appropriate current corresponding to the KG equation (see Excercise 2. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). (2π)3 2Ek (2. −k) . Then. (2.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. The KG equation. rather than as a wave function. 2. there are no longer fundamental objections to mix up space and time.14) ˜ ˜ Introducing φ(Ek . At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. (2. having the KG equation as a space-time symmetric (classical) equation. but rather the zero component of a four vector. however. ∂µ j µ = 0. 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. relativistically the density is not a scalar quantity. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. k) ≡ a(k) and φ(−Ek . The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. 2M 2M (2.13) ˜ with (in principle complex) coeﬃcients φ(k).e.11) Therefore.3). This leads to the existence of negative densities.5). j) = ↔ i φ∗ ∂0 φ. As we will see later both problems are related and have to do with the existence of particles and antiparticles. which is what Lorentz transformations do.2) is j µ = i φ∗ ∂ µ φ or (ρ. And. there are solutions with negative energy. (2. j) of the four-current j.10) ∂t which follows directly from the Schr¨dinger equation. This continuity equation can be written down o covariantly using the components (ρ. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time.Relativistic wave equations 9 i. k) + ei k·x φ(−Ek . φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. we want the most general solution. 2. just as the dependence on time was in ordinary quantum mechanics. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. ↔ ↔ (2.15) . φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek .9) They satisfy the continuity equation. −i φ∗ ∇ φ .8) (2.

19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. Another symmetry is found by complex conjugating the KG equation. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . Since a · b = a · ˜ it is easy to see that ˜ b. −x) ≡ (˜µ ).14 and 2. 2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). (2. (2π)3 2Ek (2.Relativistic wave equations 10 In Eqs 2. (2. for which we consider space inversion. changing the sign of the spatial coordinates. i.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . x) → (t. (2π)3 2Ek (2.e.16) The consequence of this is discussed in Exercise 2.6 We will explicitly discuss the example of a discrete symmetry. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes).20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k).18) (2. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. Performing some Lorentz transformation x → x′ = Λx.3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. which implies (xµ ) = (t. after restricting the energies to be positive). The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. x ∂µ ∂ µ + M 2 φ(˜) = 0. (2. φP (x) ≡ φ(˜) (P for parity).17) (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).23) . 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 (2.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. This shows how parity transforms k-modes into −k modes.

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

i (a) Show that relativistic invariance.6 Write down the mode expansion for the Lorentz transformed scalar ﬁeld φ′ (x) and show that it implies that the Lorentz transformed modes satisfy a′ (Λk) = a(k). in terms of the a(k) and b(k).e. a{µ bν} ≡ aµ bν + aν bµ . β} = 0. i. ∂t where ψ is a wave function and the nature of α and β is left open for the moment. . an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. {αi . d3 x (∂ {0 φ)∗ (∂ i} φ). Note that a{µ bν} indicates symmetrization. 2. j 0 = ψ†ψ and j = ψ † αψ. 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). Exercise 2. αj } = 2 δ ij I.15. also using the result in (a). B] = AB − BA and the anticommutator as {A. 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). We will encounter these quantities later as energy and momentum. 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 . 1 2 (b) Express now the full charge QV (t) (exercise 2. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. B} = AB + BA.1) for a complex scalar ﬁeld current (exercise 2.Relativistic wave equations 12 Exercise 2.7 To solve the problem with positive and negative energies and get a positive deﬁnite density. t). making sure that each component of ψ satisﬁes the KleinGordon equation (in terms of diﬀerential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi .2) in terms of the a(k) and b(k) using the expansion in Eq. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). and β 2 = I. Exercise 2. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. ∂ ψ(r. (c) Similarly. t) = (−i α · ∇ + m β) ψ(r.5 (a) As an introduction to parts (b) and (c).

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

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

. In the full parameter space. . for ﬁxed n. . with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. .e. ˆ ˆ A(ϕ. n) i ∂ϕ = ϕ=0 (3.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + .10) σ ˆ · n. thus.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . (3. i. it is suﬃcient that the generators commute with H. immediately sees that the Lie algebras are identical. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). They satisfy σi σj σk = i ǫijk . 2 2 2 (3. The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I.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. (3. i. 3. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. (3. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ.9) The parameter-space of SU (2). sometimes in more than one way (but this will be discussed later). in particular that AB → RAB = RA RB ). SU (2) ≃ SO(3). 2! . For instance in a function space translations. n) −→ R(ϕ. n! 2! n=0 ∞ (3. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. .e. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. n) ˆ −→ R(2π − ϕ. the above mapping corresponds to the trivial mapping of the Lie algebras. H] = 0 for g ∈ G. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). Near the identity. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones.12) One. i. also can be considered as a ﬁlled 3-sphere. Suppose we have a system described by a Hamiltonian H. thus. . however. n) π ≤ ϕ ≤ 2π.14) For a Lie group.11) Thus σx /2. [g. n) ≈ 1 + i ϕ J · n.e. 2 (3. H] = 0 for g ∈ G.15) [g. .

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

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

consisting of the Lorentz group and translations. 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 . (ii) |Λ00 | ≥ 1.25) (3.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index.g.22) (Note that xT is a row vector). (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ .21) g 20 g21 g22 g23 0 0 −1 0 g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3.Groups and their representations 18 3. 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. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . In this section we consider the Poincar´ group.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.27) (3. it is convenient to use a vector notation for the points in Minkowski space. G = 10 (3.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. σ/2 in a two-dimensional (spin 1/2) case. We considered the generators and looked for representations in ﬁnite dimensional spaces. e. e To derive some of the properties of the Lorentz group. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. det Λ = −1 is called improper). Writing x as a column vector and the metric tensor in matrix form. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.26) (3. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. From this property. (det Λ = +1 is called proper.

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

(3. P+ etc. 3 30 with η = ω and K = M (e. Summarizing. inﬁnitesimally approximated by (I + ω.39) (e. (3. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . Also this group can be divided into four parts. This is a + normal subgroup and the factor group is the Vierer group.g. The extension to the Poincar´ group is e ↑ straightforward. inf (3.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . 3. i i0 i i0 1 ijk M jk 2ǫ (3.33) These four transformations form the Vierer group (group of Klein).37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . We therefore ﬁnd (again) that there are six generators for the Lorentz group.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . ǫ). we continue using covariance arguments. K = M ).34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. In order to ﬁnd the commutation relations including the translation generators. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν .38) Explicitly.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .Groups and their representations 0 0 −1 0 0 −1 0 0 = 0 0 −1 0 0 0 0 −1 ∈ L↓ + 20 Is It (space-time inversion) (3. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . We will just require Poincar´ invariance for the generators themselves. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. Of the four parts only L↑ forms a group. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. There are six generators.g.36) (3. [M µν . a).

2 2 (3. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. ǫ) = 1 − ωαβ M αβ + i ǫα P α . The generators of the translations are the (momentum) operators P µ . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . P j ] = [P i . . P ν ] [M µν . For instance. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. P j ] = i ǫijk P k . Taking any one of the states in an irreducible representation. P j ] = i δ ij H/c2 . P ρ ] = 0.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. (3. (3. Inﬁnitesimally. [J i . P µ |p. [P µ . 3.44) (3. They just state that M µν transforms as a tensor with two Lorentz indices.46) [K i . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. . [K i . [J i . the generators J 2 and J 3 in the case of the rotation group. H] = 0. . known from non-relativistic quantum mechanics are obtained. (3. H] = [J i . = −i (g µρ P ν − g νρ P µ ) .41) 2 At this point. Explicitly. The eigenvalues of these will be the four-momentum pµ of the state. These form a group called the little group associated with that four vector. . the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . writing the generator P = (H/c.38 could have been obtained in the same way. one obtains 2 [P i . the generators M αβ no longer exclusively act in Minkowski space. We have here reinstated c. a) = Λµν P ν or U † (Λ.45) Note that the commutation relations among the generators M µν in Eq.48) . [J i .47) (3. . that commute e among themselves. . [K i . K j ] = i ǫijk K k .42) (to decide on where the inverse is. H] = i P i . the generators J± in the case of the rotation group. For instance. check that U1 U2 corresponds with Λ1 Λ2 ). as part of the set. J j ] = i ǫijk J k . leads to U (Λ.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. = pµ |p. ν (3. Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . (3. other states in that representation are obtained by the action of operators outside the maximal commuting set.43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . (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 ) in terms of the Hamiltonian and the three-momentum operators. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. a)P µ U † (Λ. a)P µ U (Λ. 3. a) = Λ−1 µ P ν . . K j ] = −i ǫijk J k /c2 .

. 0. (3.54) This will enable us to pick a suitable commuting ’spin’ operator. .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. M ǫijk . Wν ] = −i (gµα Wν − gνα Wµ ) . Massive particles: p2 = M 2 > 0. . . 0.52) (3. Pν ] = [Wµ . while the others have no physical signiﬁcance. From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. the third case represents the vacuum. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . Wα ] = [Wµ .55) commute with all generators and therefore are invariants under Poincar´ transformations. = = i exp − ǫµνρσ M µν pρ sσ |p. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). i ǫµνρσ W ρ P σ . .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . 2 exp (−isµ W µ ) |p. 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. .49) (3. which can thus be chosen spacelike. 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+ . . (3. . 2 (3. p0 > 0. .53) (3.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant.56) . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij .

62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. (3. 0.58) i.e. 0). 0. In that case the vectors ni (p) are just the space directions and W = (0. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. (0. n1 (p). The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). W i (p) = −W · ni (p)). p0 > 0.65) . s. (1. We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. 0). Having made a covariant decomposition. 0.57) (i. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 1. Together with the four momentum states thus can be labeled as |M. . 2 .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. 0. 0).Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3.64) such that in an arbitrary frame where the four vectors p. 1. 0.. i (3. (0. W j (p)] = i M ǫijk W k (p). 0. M2 (3. .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).61) one sees that S i (p) i =− W2 . More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). 3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. The commutation relations [W i (p).60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . |p3 |). (3. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). M J) with components W i = (M/2)ǫijk M jk . M2 (3. p. show that W /M can be interpreted as the intrinsic spin.e. ms . it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 1 2 (3. . Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). 1.59) (3.

The algebra of W 1 . implying continuous values for the spin (actually for W i = M S i ). λ . P µ . W 2 and λ is derived from the general commutation relations for [Wµ . 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).71) which is called the helicity. 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. 3. Thus we have P 2 . (3.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). i W 2 (p).1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. Note that angular momentum does not contribute to helicity as L · p = 0. thus.67) (3. A massless particle. Thus 2 λ(p) = J ·p .67 has a vanishing right hand side). W 1 (p)] [λ(p). σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). is characterized by its momentum and the helicity.13.73) (3. But if M = 0 one has a diﬀerent algebra (Eq. 3.Groups and their representations The above does actually include the massive case.70) as a commuting set with zero eigenvalues for P 2 and W 2 . 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 . 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. The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). The vacuum: pµ = 0.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. |p. however. (3. W 2 (p)] = = = i M 2 λ(p). 1 2 24 (3. W 2 (3. |p| (3. −i W 1 (p). W 2 (p)] [λ(p). Exercises Exercise 3. Wν ]. U (Λ. which can take any integer or half-integer value. W µ (p) = λ(p) P µ . and gives [W 1 (p). a)|0 = |0 .66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). This state is in physical applications nondegenerate and is denoted by |0 .68) (3.

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

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

j ′ ).Chapter 4 The Dirac equation 4. 0) Type II : (0.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2).1 The Lorentz group and SL(2.3) (4.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. i ǫijk B k . 2 (4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . 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 . (4. Special cases are the following representations: Type I : (j. respectively. 0.10) . C). where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. The group SL(2. B j ] = i ǫijk Ak . 2 2 (4. 2 1 (J − i K). U (ϕ) H(η) (4. similarly as the homeomorphism between SO(3) and SU (2). (4. C) is the group of complex 2 × 2 matrices with determinant one. This tells us how to ﬁnd the representations of the group.5) = 1 (J + i K). (j.6) (4.9) (4.4) (4. K = −i . They precisely correspond to the two types of representations that we have seen before. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . They will be labeled by two angular momenta corresponding to the A and B groups. Aj ] = [B i . becoming the deﬁning representations of SL(2. it also follows directly that the Lorentz group is homeomorphic with the group SL(2.1) (4. K = +i . K = i J (A = 0).7) From the considerations above. j) K = −i J (B = 0). B j ] = [Ai . 2 2 σ σ J = . With this choice of parameter-spaces the plus and minus signs are actually relevant. The matrices in SL(2. It is simply connected and forms the covering group of L↑ .

11) Considering ξ and η as spin states in the rest-frame. C) or L↑ . U σ µ aµ U −1 = σ µ ΛR aµ . This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. H ≡ (H † )−1 = H −1 ). ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. (4. i. Λ(ǫ) ∈ L+ . namely H(η) = exp(η · σ/2). there exists the matrix ǫ = iσ 2 . (4.17) that can be used to relate σ µ∗ = ǫ† σ µ ǫ = ǫ−1 σ µ ǫ and hence ˜ ¯ U ∗ = ǫ† U ǫ = ǫ† U ǫ. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). Eqs 4. σ µ ≡ (1. C).18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. This shows that M ∗ ≃ M . −σ). thus. ǫ−1 = ǫ† = ǫT = −ǫ).15) (4. (ǫ∗ = ǫ. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. σ). such that M = SM S −1 . (4. This is not true for the two-component spinors in SL(2. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. In fact. if there does not exist a unitary matrix S. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. The Lorentz transformations Λ(M ) and Λ(M ). while ±ǫη ∗ transforms as a type-I + spinor. C). One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. cannot be connected by a Lorentz transformation belonging to L↑ . one can use a boost to the frame with momentum ˆ ˆ p.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. ˜ (4. C) and. the ﬁrst equation shows that the conjugate representation with spinors χ∗ transforming according to −σ∗ /2 is equivalent with the ordinary two-dimensional representation (±ǫχ∗ transforms according to σ/2). The following relations for + rotations and boosts are useful. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.12) ξ → U ξ. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. η → U η. (4. however. ˜ = σ µ Λ B aµ . They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that y + ↑ ∗ ǫ ∈ SL(2. ξ −→ Hξ. (4.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2.1). thus.16) = σ µ Λ R aµ .13) 2M (E + M ) (exercise 4.19) . η → Hη.e. H(η) = exp(−η · σ/2).18) As already discussed for SU (2). One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 .

The representing member of the class P− is the parity or space inversion operator Is . (vector). 0) are inequivalent. 0).23) u(p. −a). ) −→ ( . a) and ˜ a = (a0 .g.The Dirac equation 29 where The matrix I2 does not belong to L↑ . C) are suitable for representing spin 1/2 particles. e. 0) of SL(2. etc. 0 χm uR H(p) .e. we have seen that in the rest frame W i (p)/M → J i . We also have seen that the representations (0. 2 ) and ( 2 . where a = (a0 . It has the same eﬀect as lowering the indices. Thus M ∗ and M are not equivalent within SL(2. This provides the easiest way of seeing ˜ what is happening. 4. m) = χm . The behavior is the same for classical quantities. but to L↑ . (4. From the deﬁnition of the representations (0. 1 ) and ( 1 . −→ −r (vector). = (4. 1 ) and ( 1 . For a particle at rest only angular momentum is important and we can choose ξ(0. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. −→ M (Λ) (4.1. but by a transformation belonging ↑ ↑ to the class P− . (scalar). m) = χm uL H(p) 0 . the well-known two-component spinor for a spin 1/2 particle. C). ǫµνρσ will change sign. Within the Lorentz group. i. (pseudoscalar). they cannot be connected by a unitary 2 2 transformation. but rather the operators W i (p) should be used. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. (axial vector). that have the correct commutation relations.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. generators. 4. must combine the repe resentations. they can be connected.17.20) 4.13. Although these operators cannot be used to label the representations. 2 2 (4. i. Taking M (Λ) = H(p). The i angular momentum operators J are represented by σ i /2. therefore. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . we obtain for the two components of u which we will refer to as chiral right and chiral left components.e. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . −→ −p (vector). Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar).22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. the aij elements of a tensor will not change sign. For the spin 1/2 representations of the Poincar´ group including parity we.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. as we have seen in section 4. m) = η(0. the boost in Eq. A parity transformation changes aµ into aµ .

29) which is an explicit realization of the (momentum space) Dirac equation.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . p (4. We will discuss another explicit realization of this algebra in the next section. Since ∂µ ∂ν is symmetric. (4. 4. The explicit realization appearing in Eq. γ } ∂µ ∂ν + M 2 ψ(x) = 0. = (4. = H −1 (p). To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). (4. 0 H 2 (p) uR uR . (4. (4. The general requirements for the γ matrices are thus easily obtained.33) deﬁne for a four vector a the contraction a = aµ γµ / . (iγ µ ∂µ − M ) ψ(x) = 0. this can be rewritten 1 µ ν {γ .24) (4. The ﬁrst indeed is solved.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). γ ν } = 2 g µν .27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.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. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. (4.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u.7. Applying (iγ µ ∂µ + M ) from the left gives (4. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. {γ µ . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. It is of a form that we also played with in Exercise 2.32) suppressing on the RHS the identity matrix in Dirac space. . 2 (4. which in general is a linear equation in pµ . From this one obtains the Cliﬀord algebra for the Dirac matrices.27 is known as the Weyl representation.28) which is a covariant (linear) ﬁrst order diﬀerential equation.

4. Relying on the Pauli exclusion principle for spin 1/2 particles. The Dirac sea forms the vacuum.g. Dirac supposed that the negative energy states were already completely ﬁlled and the exclusion principle prevents any more particles to enter the sea of negative energy states. (4. This can most easily be seen in the particle rest frame.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices.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 . This hole forms an antiparticle with the same mass. one sees that there are two positive and two negative eigenvalues. but with properties such that it can be annihilated by the particle (e.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4.36) (4.34) for the adjoint spinor ψ ≡ ψ † γ0 . From the vacuum a particle can be removed. The second problem encountered before. γ ν } = 2 g µν . E = +M (twice) and E = −M (twice). (4. the one of the negative energy states. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).3 General representations of γ matrices and Dirac spinors {γ µ . γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. The density j 0 = ψγ 0 ψ = ψ † ψ (4. 4.27). (4. This problem was solved by Dirac through the introduction of a negative energy sea. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. . where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.37) Using the explicit form of γ in the Weyl representation (see Eq. opposite charge). remains. .35) is indeed positive and j 0 can serve as a probability density.1: The Dirac sea of negative energy states and antiparticles.

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 . (γ5 )S.43) S=√ . 2 (4.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .41) (4. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. We note that ψ → Lψ and ψ → ψ L−1 . γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. is 1 1 1 (4.R.R. 0 (4. 4! (4. (γµ )S. the rotation and boost generators in Weyl representation in Eqs 4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . −→ Sψ. while L−1 γ µ L = Λµ γ ν . Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .46) (4. = S(γµ )W.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . For instance in the Weyl representation (but valid generally). S −1 . (4. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . −→ Sγµ S −1 . γ 2 ]. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. γ µ } = 0 and explicitly one has 0 1 1 .9 and 4.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.47) (4.R.45) (γ5 )W. For example.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 .40) (4.48) are projection operators. which in the case of the Weyl representation are just the upper and lower components.R. γ σ ]. =ρ ⊗1= 0 −1 3 (4. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .3) . that project out chiral right/left states. γ 0 ].

C is real and one has C −1 = C † = C T = −C c and [C. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. Some properties of C are C −1 = C † and C T = −C.55) (4.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. usually to be taken unity. (C)W. = = 0 iγ γ = −iρ ⊗ σ = −iσ 2 −iσ 2 iγ 2 γ 0 = −iρ3 ⊗ σ 2 = 0 2 0 1 2 Thus we ﬁnd back the Dirac equation. called charge conjugation.R.) and all representations connected via a real (up to an overall phase) matrix S. ψR . One has ψ c = −ψ T C −1 . In any representation a matrix C exist.54) e. parity. 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) = 0.6). T (4. The latter implies that the conjugate of a right-handed spinor.53) (4. for instance. ∂ (4. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.g. −iσ 2 0 −ǫ = .The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. such that T Cγµ C −1 = −γµ . (or W.R. is a left-handed spinor. that brings ψ → ψ c .56) (i/ − M ) ψ c (x) = 0.R. = 0 ǫ iσ 2 (4. which is again a solution of the Dirac equation.57) (4. ψ= (4. As with parity we look for a transformation. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ .59) −→ ψ c = C ψ = η ǫ ξ∗ . Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . ∂ with the solution ψ c (x) = ηc Cψ (x). x = (t. reverse the charge of the particle. 0 −ǫ 0 0 −ǫ 0 . In S. (C)S. ∂ (4. γ5 ] = 0. but it does.58) where ηc is an arbitrary (unobservable) phase.g. Explicitly. This symmetry does not change the spin 1/2 nature.R. T (4. e. (i/ − M ) ψ(x) = 0. ψ= (4. x) → x = (t.52) η ξ one can apply space inversion.

g. ∂ −iσ · ∇ −i ∂t − M (4. with E = Ep = + p2 + M 2 . In that frame is a spacelike unit vector sµ = (0. s)v(p. (2π)3 2Ek (4.68) (4.The Dirac equation 34 4. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s′ ) = −2M δss′ .1). 2M 2M (4. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. ⇔ (/ − M ) u(p) = 0. s). s) + v(k. s ) = v (p. s)u(p. v (p. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s ) = v † (p. ¯ (4.69) = − s In order to project out spin states. s)v(p. s) = u(p. 2M 2M −/ + M p v(p.60) ψ(x) = 0.70) = Ps = ˆ 0 1∓σ·s 2 2 . s) ei k·x d∗ (k. s) = v(p. v(p. s)¯(p. s′ ) = 2M δss′ .67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. (4. s′ ) = 2Ep δss′ . The ¯ ¯ solutions are normalized to u(p.. ¯ ¯ † ′ u (p. the best representation to describe particles at rest is the standard representation. s) .61) σ·p −(Ep + M ) There are also two negative energy solutions. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4.63) u(p. s)v(p. s)u(p. the spin polarization vector in the rest frame is the starting ˆ point. p (4. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. s) e−i k·x b(k.64) (4. s ) = 0. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s) and C v T (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. be obtained by a Lorentz transformation. ψ(x) = u±s (p) e−i p·x . ψ(x) = v ±s (p) ei p·x . s). s) = u(p. in which γ 0 is diagonal (see discussion of negative energy states in section 4. s)u(p. ⇔ (/ + M ) v(p) = 0. In an arbitrary frame one has s · p = 0 and s(p) can e.62) where χs are two independent (s = ±) two-component spinors. where u satisﬁes −σ · p Ep − M u(p) = 0. s) u = . the spinors satisfy C uT (p.65) (4. ¯ ′ ′ Explicit solutions in the standard representation are χs .66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. s) v = . s) = Ep + M σ ·p Ep +M χs . s)¯(p.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions.

−1/2) = √ (4. λ = +1/2) = v(p. . −p3 −p1 − i p2 1 v(p. √ 0 u(p. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . In principle massless fermions can be described by twocomponent spinors. p ψR/L ≡ PR/L ψ are independent solutions. u(p.74) . This means that in the solution space for massless fermions the chirality states. λ) and C uT (p. λ) = v(p. 4.73) Also for helicity states C uT (p. helicity states. λ = −1/2) = u(p. Note that for solutions of the massless Dirac equation /ψ = 0. We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. The chirality projection operators in Eq. λ = − 1 ) = √ E + M − E − M . λ). 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 √ . (4. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). v(p. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. u(p. v(p. +1/2) = √ 0 E +M E+M .48 replace the spin projection operators which are not deﬁned (by lack of a rest frame).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 ). 2 2 0 √ 2 √ 2 − E+M − E−M 0 . etc. Explicit examples of spinors are useful to illustrate spin eigenstates. Therefore. (4. λ) = u(p. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . λ = + 1 ) = √ + E +M − E −M . one has in standard representation: 0 E+M E+M 1 1 0 . λ = − 1 ) = √ √ v(p. while the negative helicity (λ = −1/2) is in essence lefthanded. chirality. −1/2) = √ E + M −(E + M ) 0 . +1/2) = √ u(p. = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. For instance with the z-axis as spin quantization axis. 1 u(p. λ = −1/2) = √ − E+M √ 0 0 E+M (4. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p.71) p − i p2 .72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M .

s).5 γ gymnastics and applications An overview of properties can be found in many text books. (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . (6) Tr(γ5 ) = 0. we will give one example. (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ). (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 . (8) σµν ≡ (i/2)[γµ . s)¯k (k. s′ ) ¯ u s. k ′ ) = 2 ′ s. more convenient ¯ to use the projection operators introduced earlier.g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . s′ )(γ µ )ij uj (k. It is. (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 . however.75) Lµν (k. k k 2 . etc. s′ )¯i (k ′ . s′ )) = u(k ′ .s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ .) (2) Tr(1) = 4. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν .) In order to show the use of the above relations. u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ . (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. s′ ). s)(γ ν )kl ul (k ′ . namely the calculation of the quantity 1 ∗ (¯(k. s)γ µ u(k ′ . s)γ µ u(k ′ .g µρ g νσ + g µσ g νρ ). (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. s)γ µ u(k ′ . γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. Some elementary properties are: Properties of products of γ-matrices: (1) γ µ γµ = 4 (2) γ µ γ ν γµ = −2 γ ν or γ µ /γµ = −2 a. s′ )(γ µ )ij uj (k. s′ )) . u u (4. s′ )γ µ u(k.s which appears in quantum electrodynamics calculations (see Chapter 10) for the emission of a photon from an electron changing its momentum from k to k ′ (k 2 = k ′2 = m2 ). In principle one can take a representation and just calculate the quantity u(k. s)(γ ν )kl u u s. s)γ ν u(k ′ . For this we ﬁrst have to prove (do this) that (¯(k.s′ ∗ (4.76) ul (k ′ .The Dirac equation 36 4. (Use (γ5 )2 = 1 and pull one γ5 through. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ . s′ )) (¯(k. a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b. s)¯k (k.

a 4 a · b. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.1 Prove Eq. x → x. s).The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices.4 Show that in P+ ≡ s=±1/2 u(p. 4. Show that the equality holds. of which only the traces of four and two gamma-matrices are nonzero.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . x ˜ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. (4.6 Apply space-inversion. Do this e.5 Show starting from the relation {γµ . e. 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 . H(p) = exp η·σ 2 = M +E+σ·p . Exercise 4.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. where x = (t. Excercise 4. −x) is also a solution of the Dirac equation.g.13. s) and v(p. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜).77) Exercises Excercise 4. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . i S ρσ ] = g ρµ γ σ − g σµ γ ρ . ˜ Exercise 4. s)¯(p. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. by letting them act on the four basis spinors u(p. 4 2 Exercise 4. s) u p /+M = 2M 2M both sides are projection operators. γ σ ] = σ ρσ . 0 if n is odd.g.

t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. Excercise 4. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal).e.511 MeV). Do you understand why? The calculation of the life-time is completed in chapter 11. The pion has spin 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 ). we want to conﬁne this solution to a spherical cavity with radius R.8 × 10−6 . the decay would even be forbidden. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. there is no current ﬂowing through the surface of the cavity.The Dirac equation 38 Exercise 4. ν p Note: how such parametrizations work will be explained in chapter 8.66 MeV. n It turns out that there is actually a whole family of conditions that provide conﬁnement. 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. Note: To prove this. is a stationary solution of the (free) Dirac equation.57 MeV. mµ = 105. 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 ). 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. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. Calculate the value of k for M R = 0. Calculate for now just |M |2 . namely i / ψ = eiαγ5 ψ. (b) Next. We will not pursue this chiral freedom at this point and take α = 0.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. i.5 and M → ∞. . me = 0. r). M R = 1. For such a cavity nµ = (0. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. you need to determine the corresponding conﬁnement condition for ψ. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0.

The vectorﬁeld is therefore (λ) suited to describe spin 1. λ) + ǫ(λ)∗ (k) eik·x c∗ (k. µ (2π)3 2E µ (5. In an arbitrary frame ǫµ (k) with λ = 0. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . In the restframe.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. These can describe spin 1 states in the rest-frame.Chapter 5 Vector ﬁelds and Maxwell equations 5.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.± 5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. where k = k0 = (M. ǫλ ). with ∂µ V µ = 0 (Lorentz condition). one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .1 Fields for spin 1 In section 3. (2 + M 2 )Vµ = 0. M2 (5.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. 0). there are then indeed three independent possibilities. (5.6) where Aµ is the four-vector potential (5.7) . ±(1) are obtained by boosting the restframe vectors.5) −E B2 .1) The Lorentz condition implies that k µ ǫµ (k) = 0.4) λ=0. namely ǫµ (k0 ) = (0.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. just as the two-component Dirac spinors did for spin 1/2. 39 (5. These are solutions of the Klein-Gordon equation with in addition a condition. λ) . A). −B 1 0 3 With the electromagnetic ﬁeld tensor (5.2) (5.

0).5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. transverse or Coulomb gauge). shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. (5. moree over.8) (5. One possible choice is the gauge A0 = 0 (5. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). if ∂µ Aµ = 0 for large times |t|. |k|). while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). (5. (5. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ .17) (radiation. .9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. It states the absence of magnetic charges and Faraday’s law. (5.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. i.14) of which the solutions give the Li´nard-Wiechert potentials. ∓1.15 the d’Alembertian is replaced by 2 + M 2 .15) (5.11) This equation is not aﬀected by a gauge transformation.e. AP (x) = Aµ (˜). Although the Lorenz condition is a constraint. 2Aµ = 0.16) (5. Taking the divergence of this equiation. in which case one has 2Aµ = jµ . 40 (5. For electrodynamics one has furthermore the freedom of gauge transformations. it will vanish at all times. for M = 0. Therefore if k µ = (|k|. 0. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0.12) (5.13) Hence. if in Eq. Furthermore. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . j) the Maxwell equations read ∂µ F µν = j ν . ∂µ Aµ = 0. only two independent vectors remain ǫ(±) µ (k) = (0. (5. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. As remarked above. The equation in vacuum. ˜ x Note that via parity transformation one obtains another solution. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. 0.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ.10) the electric and magnetic ﬁelds are unchanged. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. 5. in accordance with what we found in section 3. −i.

. where Aµ = 0. This occurs in the region outside the solenoid.1: The Aharonov-Bohm experiment 5. A= By Bx Br ϕ= − ˆ . but where E = B = 0. 0). but A = 0 (hence there is structure in the vacuum).20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. .0 . nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions).0 . 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 . 5.3 The electromagnetic ﬁeld and topology∗ The electric and magnetic ﬁelds E and B as appearing for instance in the Lorentz force on a moving charge. Inside solenoid: Outside solenoid: B = B z. 2 2 2 BR2 BR2 y BR2 x A= . ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. illustrated in ﬁg. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). cos ϕ.19) are gauge invariant. 0) and ϕ = (− sin ϕ. nevertheless. (5. = exp −i The phase diﬀerence between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. To be precise. This causes an interference pattern as a function of ∆x. It is however. . In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum).1.Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. ˆ B = 0. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . This phase diﬀerence is actually observed. while the electromagnetic ﬁeld Aµ is not. may look a bit akward. λ L − λ λ or ∆x = (L− λ/d)δ. F = e E + ev × B (5. sin ϕ. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. The situation.

ϕ=2π The function χ in Eq. however. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. organizing itself in tiny ﬂux tubes (Abrikosov strings). Another situation in which the topology of space can be used is in the case of superconductors. Now back to the Aharonov-Bohm experiment. 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. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . The AharonovBohm experiment shows a realization of this possibility. nonobservable phases ϕ = 2π n. This situation that A can be written as ∇χ is always true when ∇ × A = 0.e.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. implying that it must be single-valued1 .21) (ϕ being the azimuthal angle). it can be obtained from the situation A = 0 by a gauge transformation. In such a space loops with diﬀerent winding number (i. the number of times they go around the defect) cannot be continuously deformed into one another.21. 1 The occurrence of nontrivial possibilities.e. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. Therefore space outside the solenoid doesn’t care that χ is multi-valued. minimizing the space occupied by B ﬁelds. i. Therefore the gauge transformation must be uniquely deﬁned in the space one is working in. 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. forming a simple (connected) space.2: The topology of the space in the Aharonov-Bohm experiment. 5. Consider a superconductor. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. The only way for ﬂux tubes to be formed and move around without global consequences is when each ﬂux tube contains a ﬂux Φ such that ∆δ = (e/ )Φ = n · 2π. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . A = ∇χ with χ= BR2 ϕ 2 (5. If this would be be happening in empty space one would be in trouble.

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

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

e. ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ .13) (6. λ=0 (6. (6.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 (R3 . = φ(x) + δφ(x) (6.18) 1 Taking a functional derivative. which in classical mechanics vanishes at the boundary) does not play a role. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ.t1 ) (6. t2 ).6 can be rewritten as δS = . ˙ (6. (R3 . . x) ≡ p x − L . giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ . .12) Here R indicates a space-time volume bounded by (R3 . . with ∆φ(x) = δφ(x) + (∂µ φ)δx . ∂µ φ. ∂µ φ(x)). indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). We will come back to it in a bit more formal way in section 9. . H is a conserved quantity. indicated with δF [φ]/δφ should pose no problems. t1 ) and R3 . R (6. The hamiltonian H is deﬁned by H(p.14) (6.2.15) d4 x′ L (φ′ . + ∂L ∆x − H δτ ∂x ˙ t2 .t2 ) d3 x . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . their derivatives and possibly on the position. δ(∂µ φ) (6.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). . Consider a lagrangian density L which depends on these functions. ∂µ φ(x)) = R d4 x L (φ(x). . . t1 (6. In general any continuous transformation. x′ ) − d4 x L (φ. 6. In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). but which in 3 dimensions includes conserved quantities related to rotations and boosts.9) which is done because the ﬁrst term (multiplying ∆x. . i. x).11) λ=0 of which p and H are the simplest examples related to space and time translations. − d3 x . Variations in the action can come from the coordinates or the ﬁelds. L (φ(x). ∂µ φ′ . think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). The resulting variation of the action is δS = R µ ′ = xµ + δxµ . ˙ x The second term in Eq. . dσµ .17) Furthermore the variations δφ and δ∂µ φ contribute to δS.

The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities.20) (6.19) δ(∂µ φ) δφ 6. 2 ∗ (6. 2 (6. (2 + M )φ (x) = 0.25) (6.26) The Dirac ﬁeld The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. ∂ (6. leading to (2 + M 2 )φ(x) = 0.24) (6. (6. leading to the Euler-Lagrange equations. The integrand of the ﬁrst term must vanish.23) (6.21) which diﬀer only by surface terms. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.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.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). 2 ψ = 0. One easily obtains (2 + M 2 )φ(x) = 0.29) . (6. Using the variations in ψ (in the symmetric form). the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found.2 Lagrangians for spin 0.27) (6.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration). δL δL ∂µ = . (6. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. the second term is irrelevant.

Peshkin and Schroeder or Exercise 12.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. (6. 2 αβ = −βα.7). L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . ψL ≡ (ψL )c = C ψL = 0 (6. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR .g. 2 (6. . but note the factor 1/2 as compared to the Dirac lagrangian. and thus (β ∗ α)∗ = α∗ β = −βα∗ . (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL .g. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).31) showing e. it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left ﬁelds ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors.38) (6. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . (6. The reasons for Grassmann variables will become clear in the next chapter. The Majorana case is in fact more general.32) There exists another possibility to write down a mass term with only one kind of ﬁelds.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. that the lagrangian separates into two independent parts for M = 0. ∂ ∂ 2 2 (6. or equivalently separate the ﬁeld into right. the mass term in the Dirac lagrangian 6. 1 M η † ǫη ∗ − M ∗ η T ǫη .31 is given by LM (Dirac) = −M ξ † η + η † ξ . We note that of which the left part coincides .36) (6. Using the twospinors ξ and η. which comes because we in essence use ’real’ spinors.and lefthanded ones. 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 .39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . 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.34) (6. namely2 LM (Majorana) = + (6.

which is the space-integral over the zero-component of a conserved current. δ(∂µ φ) For continuous transformations. the presence of a symmetry that leaves the lagrangian invariant. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). δ(∂µ φ) (6. Returning to δS the surface term is rewritten to δS = R d4 x . we consider the second term in Eq.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . these conserved quantities become operators. (6. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . . Q= d3 x J 0 (x. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. (6.. As an example for the classical ﬁeld. These do not aﬀect the dynamics and will give rise to conserved currents. which in a consistent picture precisely generate the symmetries. In particular when the lagrangian is invariant under symmetries. Q(t1 ) = Q(t2 ). (6.. + ∂R dσµ dσµ ∂R ≡ where d x . ∂µ J µ (x) = 0.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. one can consider the variations at the initial or ﬁnal surface.47) .+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . 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 ) = 0.3 Symmetries and conserved (Noether) currents Next.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . δ(∂µ φ) (6.σ1 with σ = (R3 . discussed in the next chapter.40) 6. (6.41) (6. (6. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. 6. t).18. . requires the presence of a conserved quantity. In this case.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 .43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). t).44) δL ∆φ + Θµν (x)δxν .

Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . Summarizing. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . .56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν .g. (6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. (6. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). (6. In the next section we will see the quantity show up as the charge operator. we obtain (since δxµ = 0). ↔ (6.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.3) j µ = i e φ∗ ∂µ φ.52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). 1 (6. (6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν .49) These currents are conserved as discussed already in chapter 2.51) (6.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. 2 (6. = 0. ˙ Q = d3 x j 0 (x.53) These are the energy and momentum. 6. e.50) (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. 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.54) (6. The integral over the zero-component. H = d3 x Θ00 (x). t) is the conserved charge (Q = 0). including e the space-time translations and the Lorentz transformations. From Noether’s theorem one sees that the current Θµν ǫν is conserved.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.

φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). In general this tensor is not symmetric. (6.70) .g.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. 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 . one has a tensor that satisﬁes (exercise 6.66) In section 6.67) i.64) (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). in which the U (1) transformation involves an angle e Λ. • Vector ﬁeld Aµ : −i Sρσ = aρσ .e. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. Deﬁning T µν = Θµν − ∂ρ Gρµν . where H ρµν is the quantity appearing in M ρµν .58) (6. is not invariant under local gauge transformations.69) i eΛ(x) φ (x). however. 2 (6.68) (6.6) T µν = T νµ . the angle of the transformation depends on the space-time point x. (6. (6. −i eΛ(x) ∗ i eΛ(x) (6.3). independent of x.61) A ﬁnal note concernes the symmetry properties of Θµν . φ(x). ∂µ T µν (6. (6.60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ .63) µν = ∂µ Θ . (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).65) M µρσ = T µρ xσ − T µσ xρ . • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. 6. Any lagrangian containing derivatives. e.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. ∂µ φ(x) + i e ∂µ Λ(x) e (6.59) (6.3 we have seen global transformations or gauge transformations of the ﬁrst kind. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.

i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. 1 L (φ.72) (6.73) (6. (6. 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν . As an example consider the Dirac lagrangian. ∂ 2 4 (6.78) . (6.77) involving the interaction of the charge (ρ = j 0 ) and three-current density (j) with the electric potential (φ = A0 ) and the vector potential (A). ∂µ φ) =⇒ L (φ. −e j µ Aµ = −e ρφ + e j · A.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 . A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x). 4 (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. 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. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector ﬁelds to the original lagrangian therefore produces a gauge invariant lagrangian. D ∂ A (6.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. (6. Dµ φ) − Fµν F µν . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).76) We note ﬁrst of all that the coupling of the Dirac ﬁeld (electron) to the vector ﬁeld (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ .71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ.

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

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

The bilinear product for the conserved quantities. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. (7. An immediate generalization for ﬁelds can be obtained by considering them as coordinates.11) found through Noether’s theorem. [q. p] = i. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. H]P dt and dO = [O. B]P = dB dA dA dB − . is the Poisson bracket [A. q) = ˙ 1 mq 2 − V (q). q) also considered in the previous chapter. (7.3) For variations δO(p. t). however. namely [q. Q]P . p]P = 1.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). p]P .Chapter 7 Quantization of ﬁelds 7. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. q) = p q − L(q. ˙ 2 (7. labeled by the position. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. starting from the lagrangian L(q. Further. dλ Examples are dO = [O. 6.2) H(p.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. (7. 1 qx (t) = 3 d3 x φ(x. ˙ L(q. dq dp dq dp (7. in our example p = mq. q(q.5) ∆ x ∆3 x 54 . there are conserved ˙ ˙ quantities Q (Eq.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t).

py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. albeit with ’sharp’ functions. t).10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν .6) ∆3 x L x (qx . Π(y. t)] = [Π(x. a) φi (x)U (Λ.14) In fact. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. As an example.2. ∆φ. ˙ δ φ(x) (7. φ(y. for the scalar ﬁeld theory.11) (7. ∂µ φ(x)) = ˙ d3 x L (φ.7) d3 x L (φ(x). formally. (7. discussed in the section 6. t)] = i δ 3 (x − y).8) (7. a) = Rj (Λ−1 ) φj (Λ−1 x − a). Π(y. Sometimes it is useful to keep in mind that. i U † (Λ. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . (7. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . Φ(f ) = d4 x φ(x) f (x). 2 2 2 (7. • Poincar´ invariance e The full variation of the ﬁelds. t). ∇φ) (7.12) (7. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x).13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. ∂ qx ˙ ∂ qx ˙ (7. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).Quantization of ﬁelds 55 etc. qx+∆x ) . t)] = 0. for symmetry transformations. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). 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.16) . φ. The essential conditions to consistently quantize a theory then are the following. t). the discretization procedure above is an explicit example. (7. qx .15) with furthermore the relations [φ(x.

sometimes referred to as second quantization. Mµν ] = i i∂µ φi (x). ω a† a + 2 2 ω i − [Q.23) The hamiltonian in this case can be expressed in the number operator N = a† a. [φ(x). a† ] = 0. Pµ ] = [φ (x).17) (Sµν )i φj (x). a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . a† ] = a† . a† ] = 1. P ] = 0 (7. a] = [a† . j (7. and [N.4. (7. The measurements cannot inﬂuence each other or [Φ(f ). • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). [Q. P ] = i. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ .21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.Quantization of ﬁelds 56 or if U (Λ. Q] = [P. 2 i ωµν (S µν )i .18) with S µν given in 6. (7.24) It is straightforward to ﬁnd the commutation relations between N and a and a† . Microscopic causality implies local commutativity.19) 7.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q.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. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the ﬁelds through Noether’s theorem. Φ(g)] = 0. φ(y)] = 0 if (x − y)2 < 0.25) . P ] 2 (7. j 2 one has [φi (x). [a. [N.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. 2 2 (7. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . a] = −a. (7.

the real scalar ﬁeld φ(x) becomes d3 k (7.e. Note that we will often write a(k) or a† (k).29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. Π(x )]x0 =x′0 = 0.e. and (7.26) a(k) e−i k·x + a† (k) ei k·x . which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. 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. (n − 1) a|n † i.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).30) [φ(x). a(k ′ )] = [a† (k). The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .33) . i.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. but one needs to realize that in that case k 0 = Ek = k2 + M 2 . [φ(x). φ(x )]x0 =x′0 = [Π(x).28) (7. 7. and we see that a state |0 must exist for which N |0 = a|0 = 0. 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 . which represents the ’number of particles’ with momentum k. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. (2π)3 2Ek (7. is equivalent with [a(k). a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).like’ commutation relations between a(k) and a† (k ′ ). † √ a and a act as raising and lowering operators.32) (7.g. e. (7. a† (k ′ )] = 0. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n.

(2π)3 2Ek d3 k a† (k) ei k·x . For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). (2π)3 (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). .40) This procedure is known as normal ordering. (2π)3 2Ek (7.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. This term will be adressed below. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . |k = a† (k)|0 . which now contains an inﬁnite number of oscillators.46) (7.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). k 0 = Ek = k2 + M 2 ). i. n1 ! n2 ! (7. a(k)|0 = 0. . (2π)3 2Ek (7. d3 k |k k|. (2π)3 2Ek (7.41) (7. is solved by subtracting it as an (inﬁnite) constant. (2π)3 2Ek (7. deﬁning particle states |k = |k (with positive energy.35) where the vacuum contribution disappears because of rotational symmetry. |0 .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).34) where V = (2π)3 δ 3 (0) is the space-volume. .e. 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).45) (7. d3 k a(k) e−i k·x .37) (7.44) p|φ− (x)|0 = ei p·x . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).Quantization of ﬁelds 58 where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy. . and multiparticle states |(k1 )n1 (k2 )n2 .39) The problem with the vacuum or zero-point energy. writing all annihilation operators to the right of the creation operators.42) (7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. The rest of the states are then obtained from the groundstate via the creation operator.

51) (7. (2π)3 2Ek ≡ i∆− (x − y).49) (7.52) The result for the invariant commutator function is [φ(x). (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 (7.50) or as integrals over d4 k. 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 ). d3 k ei k·(x−y) = −i∆+ (y − x). = − (2π)3 2Ek (7.53) t=0 . 2π (7. (7. In order to calculate [φ(x).54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . x) = 0 and hence ∆(x) = 0 for x2 < 0. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . i.56) = −δ 3 (x). (iii) ∆(0. φ+ (y)] d3 k e−i k·(x−y) . The last item to be checked for the scalar ﬁeld are the causality condition. φ− (y)] ≡ i∆+ (x − y). φ(y)] consider µν [φ+ (x). that they satisfy the required commutation relations in Eqs (7.18).47) ′ d3 k ′ e−i k·x+i k ·y [a(k). (2π)4 (7. d3 k = (2π)3 2Ek = (7.e.48) (7. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. ∆(x) = − ǫ(x0 ) δ(x2 ).55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.17) and (7.

66) . (7. which is equivalent to the relations (only nonzero ones) [a(k).58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x).63) i. we will consider it as a repetition of the quantization procedure. (2π)3 2Ek (7.e.61) (7.57) (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. φ† (y)] = i∆(x − y). ↔ (7.62) (7.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) . The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) . 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 (7.60) = i φ∗ ∂µ φ. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.Quantization of ﬁelds 60 7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). = ∂{µ φ∗ ∂ν} φ − L gµν . extending it with the charge operator and the introduction of particle and antiparticle operators. 2 From the lagrangian density (7.59) (7.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.65) 7. (2π)3 2Ek (7. a† (k ′ )] = [b(k).

s)u† (k.70) where the last line is obtained by using the Dirac equation. ↔ i ψγµ ∂ν ψ − 2 (7. s) ei k·(x−y) x0 =y 0 . The solution is the introduction of anticommutation relations.71) (7. s)ei k·x + d(k. (2π)3 2Ek (7.67) i ψ / ψ − M ψψ gµν . (2π)3 2Ek d3 k b† (k. (7.75) (7.73) (7.78) (7.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . † {ψi (x). i. (2π)3 2Ek (7.e. s)v(k. s)e−i k·x . s)u(k. (7.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .74) (7. s)¯(k. u v (2π)3 2Ek (7. s′ )} = {d(k. s)e−i k·x + d† (k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. s)ei k·x . b† (k ′ . The quantized ﬁelds are written ψ(x) = s d3 k b(k. interchanging creation and annihilation operators.77) Note that achieving normal ordering. . ∂ 2 ↔ (7.69) (7. s).Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. {b(k. s)¯(k. 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. d† (k ′ . s). 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) .68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). jµ Θµν = = ψγµ ψ. s)vj (k.

∂ (7. one has † {ψi (x). (7. ˙ δ A0 δL = F i0 = E i . ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. 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). 4 Π0 Πi δL = 0. 7. we would have obtained [ψi (x). but has the problem of being noncovariant.86) (7.82) 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. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. ψ j (y)] = (i/x + M )ij i∆1 (x − y).88) . as the vanishing of Π0 induces a constraint.6 The electromagnetic ﬁeld 1 L = − Fµν F µν . It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. 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). (7. −λ(∂ρ Aρ ).83) From the lagrangian density the canonical momenta are = = (7. ψ j (y)} at arbitrary times one has {ψi (x).85) which reﬂects the gauge freedom. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. 4 2 This gives the equations of motion discussed before. ˙ δ Ai (7.80) 3 = δ 3 (x − y) δij . ∂ (7. For the scalar combination {ψi (x).

which reproduce the Hamilton equations. a longitudinal photon and two transverse photons. Aν (y)]x0 =y0 = i gµν δ 3 (x − y).93) which does exhibit problems with the time-like photon. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. λ)|A = 0. (d) Use Poisson brackets to calculate q and p. (7. (7. 0. c† (k ′ . In fact the commutation relations imply ˙ [Aµ (x). containing a time-like photon. and are equivalent with [c(k. 0) . q) that dO/dλ = [O. µ λ=0 (7. Q]P . i. λ). Πν (y)]x0 =y0 = i gµν δ 3 (x − y). 2. and [ℓi .e. Aν (y)] = [Πµ (x).1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). λ)e−i k·x + c† (k. 0. 0)c(k. These problems are solved by the Lorentz constraint between physical states given above. 7. for which it is suﬃcient that ∂µ Aµ |A = 0.92) 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. (b) Proof for a quantity O(p. λ)c(k. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. (7.95) Exercises Exercise 7.Quantization of ﬁelds 63 for physical states |A and |B . [ℓi . rj ]P . λ)ei k·x . what are [ri . pj ]P .3) satisfy the conditions of a Lie algebra. j = 1. µ λ=0 (7. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. Πν (y)] = 0. where Q is the conserved quantity in Eq.11. pj ]P . ˙ ˙ . i. ′ (7.94) Choosing k µ = (|k 3 |. The canonical equal time commutation relations are [Aµ (x). λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). λ) − c† (k. 6.89) with four independent vectors ǫµ . The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. It gives 3 k µ ǫ(λ) (k)c(k. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k.91) (7.e.

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

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

−r). parity (P).1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x). r) −→ xµ ≡ xµ = (t. (iγ µ ∂µ − M ) ψ(x) = 0. starting with the Dirac equation for ψ(x). 8. Both ψ p and ψ satisfy the Dirac equation. 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 . x (8. ˜ (8. T and C for classical quantities is shown in the table 1.1: The behavior of classical quantities under P.Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries.2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. Table 8. writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). The consequences of P. T. time reversal (T) and charge conjugation (C). We can determine A.1 Parity xµ = (t.

iγ µ† ∂µ − M ψ(˜) = 0. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. ˜ −v(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. if helicity λ is used instead of the z-component of the spin m. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. . x (8. λ)u(k.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). one can also consider the scalar ﬁeld and vector ﬁelds. γ x (8.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. −λ)u(k. m) γ0 v(k. m). op P (8. λ)γ0 u(k.7) = λ d k x x ηP b(k. (2π)3 2Ek (8. m) = = ˜ u(k.6) (check this for the standard representation.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.9) (8. the above operation reverses the sign of λ). m). λ)v(k. λ) e−ik·˜ + d† (k. λ) eik·x .5) (8. −λ)v(k.10) i. λ) P −1 = −η ∗ d(k. (i˜ µ ∂µ − M ) ψ(˜) = 0. λ) Pop = ηP b(k. λ) e−ik·x − d† (k. −λ) e−ik·x − d† (k. λ)γ0 v(k. ˜ Pop d(k. x The latter behavior of the vector ﬁeld will be discussed further below. λ)v(k.3) (8. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). In the same way as the Fermion ﬁeld.e. 8. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. −λ).13) ψ(x) −→ ψ c (x) = ηC C ψ (x).4) (8.11) (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ). −λ) e−ik·x − d† (k. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). λ)u(k.

As time reversal will transform ’bra’ into ’ket’. λ)C v T (k. Top |φ = φt | = (|φt )∗ . (8.e. λ) e−ik·x + b† (k. m). x iγ µT ∂µ − M ψ(−˜) = 0. λ). λ) −1 Cop = ∗ ηC b(k. λ).19) Cop d(k. x x (8. x (8. λ) eik·x .14) (8. λ) e−ik·x + b† (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. we start with the complex conjugated x Dirac equation for ψ. antilinear operator is anti-unitary if A† = A−1 .18) (8. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜).Discrete symmetries 68 the latter being also a solution of the Dirac equation. it is antilinear1 . The (time-reversed) Dirac equation becomes. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . r) −→ −˜µ ≡ −xµ = (−t. x (8. λ)u(k. . ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. i.17) 3 = λ This shows that −1 Cop b(k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ .21) where A is a 4 × 4 matrix acting in the spinor space. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. r). to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. λ) Cop = ηC d(k. u(k. as discussed in section 4). m). x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. (2π)3 2Ek (8. λ)v(k. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. 8. λ)C uT (k.16) d k ηC d(k.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). Norm conservation requires Top to be anti-unitary2 . (8. m) ¯ C v T (k. m) ¯ = = v(k. The same relations hold for helicity states.15) (where one must be aware of the choice of spinors made in the expansion.

26) d k x x ηT b(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(−˜).31) (8. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. (8. λ |¯. −p.33) (8. λ)u∗ (k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At .Discrete symmetries 69 Table 8. there are 16 independent of these bilinear combinations. λ) eik·x + d† (k. λ) i γ5 C v(k.28) (8. λ| ¯ C |¯. op T (8. a state |a. λ a P |a. −λ a T a. λ)v ∗ (k. ˜ = 0|b(k)|A 8. λ)iγ5 Cv(k. p. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ)u∗ (k. λ a |a. λ). λ) e−ik·x (2π)3 2Ek (8. The result is = = λ (8.29) In table 2 the behavior of particle states under the various transformations has been summarized. λ) T −1 = η ∗ d(k.34) . λ)v ∗ (k. p. λ).2: The transformation properties of physical states for particles (a) and antiparticles (¯).25) (check this for the standard representation). λ)iγ5 Cu(k. −λ |¯. λ) eik·˜ + d† (k. −p. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. λ| a. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k.30) (8.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k.24) (8. −p. p. ˜ Top d(k. λ). p. ∗ ˜ v (k. x (8.32) (8. λ) Top = ηT b(k. λ). λ)u∗ (k. Since there are 16 independent 4 × 4 matrices. λ)v ∗ (k. −p.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k. λ) eik·x + d† (k. λ) eik·x + d† (k.

g. As an example consider the vector current for a point fermion. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . (8. (q 2 ). Note that the lagrangian density L (x) → L (−x) under Θ. of which the expectation value between momentum eigenstates can be simply found. 8. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).5 Form factors Currents play an important role in ﬁeld theory. in this case only q 2 = (p′ − p)2 . p′ . called form factors. as well as under the combined operation Θ = PCT (see Table 3). ¯ In general the expectation value between momentum states can be more complicated.36) (8. p) can be built from any combination of Dirac matrices (1. and T. ¯ ′ ′ (8.39) where Γµ (p′ .35) The x-dependence can be accounted for straightforwardly using translation invariance. 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 . p′ |V µ (0)|p = u(p′ )γ µ u(p). (8.. For instance for nucleons one has Γµ (p′ . In many applications the expectation values of currents are needed. p)u(p) e−i (p−p )·x . There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ). e. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . s′ |V µ (x)|p.. p′ |V µ (x)|p = u(p′ )Γµ (p′ .40) where the coeﬃcients are functions of invariants constructed out of the momenta. The 16 combinations of Dirac matrices appearing above are linearly independent. F. C. for the vector current V µ (x).37) (8.38) (8. (q 2 ). ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. σ µν . Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x).3: The behavior of the independent bilinear combinations of fermi ﬁelds under P.41) 2M 2M .Discrete symmetries 70 Table 8. s . γ5 γ µ or γ5 ). γ µ . 2M (8. γ ν ]. C. F.

p′ )γ0 = Γµ (p′ . µ p ˜ (8. p) that Γµ (p′ . p)u(˜) ˜ x p ¯ p p ˜ p (8. p) = (i γ5 C)Γ∗ (˜′ . p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ .42) where q = p′ − p. p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. p). Therefore hermiticity implies for the structure of Γµ (p′ . p′ )u(p′ ) = u† (p′ )Γµ† (p.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ .43) = ei q·x u(p′ )γ0 Γµ (˜′ . • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ .45) Exercises Exercise 8.1 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 ) + ′ .g.g. the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). e. p)γ0 . p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . e. p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p)γ0 u(p). p) that Γµ (p′ . p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ .Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. p) that γ0 Γµ† (p. or relations that follow from the equations of motion. or relations based on hermiticity of the operator or P . ¯ 2M (8. ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . p′ )γ0 u(p). p) = γ0 Γµ (˜′ . • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . p)(iγ5 C)u(p). C and T invariance. p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . p)(i γ5 C). ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ .

Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . Exercise 8. −|q|/2). (c) Show that hermiticity of the current requires that F1 en F2 are real.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. 0. (b) Show that in the interaction with the electromagnetic ﬁeld. |q|/2) with E 2 = M 2 + |q|2 /4 and express them in terms of the Sachs form factors depending on Q2 = −q 2 . 4M 2 = F1 + F2 . 0. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M .3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 0. 0. 0. (c) Show that if a term of this form would exist. σµν q ν F3 (q 2 ) 2M . (b) Show that such a term is also not allowed by time-reversal symmetry. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). p′ = (E. p = (E. |q|). Exercise 8. GE GM = F1 − Q2 F2 . 0.

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

t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . 2π (9..13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.8) Choose t as the starting point with |q = |q. B] + . Then qj+1 .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 ′ . ˙ 2π 1 1 1 [A.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 . [A. Combining the two terms gives qj+1 . (9. eA eB = eC with C = A+B+ (9. t′ |q.q . The hamiltonian is an operator H = H(P.14) . B]] + [[A. Q) = K(P ) + V (Q) = (P 2 /2M ) + V (Q) one can split e−i H∆τ ≈ e−i K∆τ e−i V ∆τ . ′ (9. which is a hamiltonian function in which the operators are replaced by real-numbered variables. . (9. .11) (9.9) 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 ′ . qj )]) . n |q (9. 2 12 12 (9. t = q ′ |e−i H(t −t) |q . with the correction1 being of order (∆τ )2 . |q1 q1 |e−i H∆τ |q ..16) this. H(P. t′ |q.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . tj+1 |qj . use the Campbell-Baker-Hausdorﬀ formula. B]. The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . . Q) expressed in terms of the operators P and Q. B] + [A. . 2π (9. qj ).12) where the transformation between coordinate and momentum space involves q|p = ei p.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . tj+1 |qj . These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|. QS |q ′ ≡ q ′ |q ′ .

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

A useful property of functional integration is the translation invariance. Dα F [α] = Dα F [α + β].n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. (9. 2 (9.21) again a multidimensional integral.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.27) (9. Consider the gaussian functional as an example.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. N (9. 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) (9.25) (see Exercises). 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. the following integrals can be derived for a real symmetric or complex hermitean kernel K. (9.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . with Dα F [α] = n dαn N F (αn ) (9.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration. Note that procedure (i) is an example of the more general procedure under (ii).28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .23) Having deﬁned the Gaussian integral. y) = ∗ m. .24) (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.

θη = −ηθ. dx dy θ∗ (x) K(x. (9.31) δ δ α = dy α(y) = dy δ(x − y) = 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.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ .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.32) (9. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. αy = δxy . y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 .34) (9. e.33) (9. F [θ. In the expansion of the exponential (from second to third line).38) .29) (9. y) θ(y) . the Gaussian-type functionals. δα(x) or in discretized form δ 1 ∂ ∂ with . α(y) = δ(x − y). −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . however. αn = δmn . δα(x) δα(x) δ αKβ = dy K(x. n θn fn (x) dx θ∗ (x) θ(x) . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . i. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). (9.e. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. In principle the deﬁnitions of functionals is the same. θ2 = 0. Examples of functional diﬀerentiation are (9.g.30) (9. (9. products of the same pair appear. δα(x) (9.35) For applications to fermion ﬁelds. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. which vanish.

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

Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. T with q. for g. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. t′ |q.e. t) φ∗ (Q) e+i En T . T ′ |Q. q) → L(q. L(q. q)] ˙ (9. q) + J(t) · q. T < t < t′ < T ′ . it is possible to switch on an interaction.s. φn (q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). q . t q. i. ˙ J (9.5 The generating functional for time ordered products By introducing a source-term. say. n (9. 9. q) + J(τ )q] .62) p Dq D exp i 2π t dτ [pq − H(p. t = ei Ht |q . t|Q. Consider. the creation of an electron (think of a radio-tube. which in turn are embedded between an early time T and a future time T ′ . Considering the source to be present between times t and t′ . Before and after these processes there is only the vacuum or ground-state |0 . T = n q. t|n n|Q. making the electron) and the absorption of an electron (think of a detector). i. one has q ′ . T .1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. t′ q ′ .e. t|n = q|n e−iEn t with q|n the time-independent wave function. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. t |q. a set |n of physical eigenstates. furthermore. T ′ |q ′ . T = n J t′ = = Dq exp i t dτ [L(q.65) J dq ′ dq Q′ .66) . t|Q. t′ |q. The Heisenberg state |q. q)] . t is related to the Schr¨dinger state |q by |q. t = = ′ ′ V = q . t Q′ .63) for plane waves. harmonic oscillator.64) (9. ˙ ˙ physically pictured as. for the physical states o q.

T = φ0 (q. 9. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. dq dτ + J(τ )q(τ ) . t′ |q. t 0 0out |0in J . . . J (9. t) (9.67) = = φ0 (q. The importance of Z[J] is that the time ordered product of operators can be obtained from it.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 ′ ).6 Euclidean formulation The above expression (Eq. t). t|Q. Q(tn )|0 .Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. t′ ) q ′ . Since we have (neglecting multiplicative factors). (9. .70) for the generating functional is in fact ill-deﬁned.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. δJ(tn ) hence the name generating functional.71) 9. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. ˜ t′ → ∞. T → i ∞ and T ′ → −i ∞. J=0 (9. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . thus T →i∞ lim q. Better is the use ˜ of imaginary time t = −it. . .70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .68) (9. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. = (i)n 0|T Q(t1 ) . T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. T ′ |Q.

n fm (x) Kmn fn (y). 1 = √ . det K Exercise 9. ensuring convergence for the functional integral ZE [J].75) = − LE q.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. ZE [J] = = where LE q. . dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. ˜ dt (9. τ dq ˜ dt ≡ −L q.73) d˜ LE q. (9. .76) which is a positive deﬁnite quantity. e. + V (q). δJ(tn ) = (i)n J=0 1 δ n ZE [J] . i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. The diﬀerentiations in Z[J] and ZE [J] are related. when L q.74) − V (q) 2 (9. ˜ ˜ ZE [J] δJ(t1 ) . where Kp are the eigenvalues of the matrix Kmn in K(x. . det K 1 . dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). y) = ∗ m. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp .g.72) (9.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. . As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. δJ(tn ) J = 0 ˜ t = it (9.and Grassmann-valued functions. Exercises Exercise 9. i dq . complex. δ n Z[J] 1 Z[J] δJ(t1 ) . . for Grassmann valued functions.2 Check the examples of functional derivation for real-. (9.

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

φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . . . implying at the level of four vectors in coordinate and momentum space (when k · x ≡ kE · xE ) for instance x4 = i x0 = i t x · x = (x0 )2 − d4 xE = i d4 x L = 1 2 3 i 2 i=1 (x ) k 4 = −i k 0 =− 4 µ 2 µ=1 (x ) k · k = (k 0 )2 − d4 kE = −i d4 k 3 i 2 i=1 (k ) =− 4 µ 2 µ=1 (k ) ∂µ φ∂ µ φ − 1 2 M 2 φ2 LE = 1 2 µ (∂E φ)2 + 1 2 M 2 φ2 Furthermore we ﬁnd the n-point Green functions 0|T φ(x1 ) . Π) (10. . (10.e. x′ . .1 Generating functionals for free scalar ﬁelds The generating functional for quantum ﬁelds is a generalization of the results in the previous section to a system with more degrees of freedom. .Chapter 10 Feynman diagrams for scattering amplitudes 10. 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. ∂µ φ) + Jφ] (10.3) The Euclidean formulation is as before. i. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. ∂µ φ) − Jφ] (10. . δJ(xn ) 85 ≡ G(n) (x1 . . t′ |φ.2) = ∞ −∞ d4 xE [LE (φ. . φ(x. xn ).4) J=0 .

.5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).e. . 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. (10.6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y).Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .13) . . . As discussed in the previous section (exercise 9. .3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . . with Z0 [0] = 1.8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. (2π)4 (k 0 + iǫ)2 − E 2 (10. 2 2 → (10. J(xn ). xn ) J(x1 ) .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). xn ) = x2 xn .10) J=0 In order to determine ∆F . Furthermore. . . These Green functions appear in the expansion Z[J] = n in n! d4 x1 .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. d4 xn G(n) (x1 . .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. .9) i. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. (10. . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. . (10.

i. φ+ (0)] = (2π)3 2E are also shown in exercise (7. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . ∆(x) = − d4 k e−i k·x .18) Thus we see various solutions. t) = 0 for t < 0. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . 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.16) C where C is a closed contour in the complex k 0 -plane. t) = 0 for t > 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). 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 ﬁrst possibility is to consider the well-deﬁned Euclidean formulation.22) (10.17) d4 k e−i k·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). φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).14) They are solutions of the inhomogeneous equation.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . (2π)4 k 2 − M 2 (10. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. (2π)4 k 2 − M 2 (10.15) e−i k·x d4 k . ∆F = ∆C . (10.e.4).21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . 4 k 2 − M 2 + iǫ (2π) (10. the advanced Green’s function ∆A (x) = satisfying ∆A (x. .20) = exp − 2 where or (see ﬁg. µ 1 1 • Firstly. (2π)4 (k 0 − iǫ)2 − E 2 (10.

x2 . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . x3 .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).1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . we have G0 (x1 . the same result as above. 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.Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. We ﬁnd (see also section 7. = θ(x0 − y 0 ) 0|[φ+ (x).25) x1 x2 x3 = x4 + + . because this is the only quantity entering Z0 [J]. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. or diagrammatically (4) (10. . . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] .

x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.34) Introducing a vertex point to which four propagators are connected.e.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. x2 ) = G0 (x1 . L (φ) = L0 (φ) + LI (φ). (10.32) (10.Feynman diagrams for scattering amplitudes 89 10.30) and in zeroth order G(0. x3 . the generating functional can be written.35) z . 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.28) 1 exp − i 2 = exp 1 2 d4 x d4 y exp i δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)] φ=0 .g ) (x1 . i.27) When interactions are present.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 .g 0 ) = G0 = 1.31) G(2. (10.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. x2 ) = i ∆F (x1 − x2 ) = G(4. Z[J] = = Dφ exp i exp i d4 z LI (10. x4 ) = G0 (x1 . x2 . x2 . (10. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. (2) (0) (10. x3 .g ) (x1 . ≡ −i g d4 z (10.

. This remains also true for the interacting case. . . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . J(xn ). . . . . (10. did not contain parts that could be written as products of simpler Green’s functions. In the free case we have seen that the exponent of Z[J] contained all essential information. . . d4 xn G(n) (x1 . . (10. (10.38) The result for the 4-points Green’s function is left as an exercise. x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . i.36) J J from which one obtains G(0.g 1 ) = 1 8 z 1 8 .41) in terms of socalled connected Green’s functions. 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]. . . which also was the only Green’s function for which the diagram was connected.39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. xn ) J(x1 ) . . . .e. δJ(xn ) . (10. J=0 (10. . .37) G(2.g ) (x1 . .40) in−1 n! d4 x1 . . . J(xn ) c (10. Diagrammatically this exponent only contained the two-point function. (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]. . .42) x1 G(n) (x1 . . xn ) J(x1 ) . . Z[J] = n in n! d4 x1 . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . d4 xn G(n) (x1 . .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . G(n) (x1 .

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

. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). one ﬁnds for the connected 4-point Green’s function at order g. . this will be translated to the action on ﬁelds. (2) (10. x2 . considered explicitly. (10. Next. in = |α.e. out|p′ . in ⇔ β. (10. Sβα ≡ β. in = |p1 . p.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. in into ﬁnal state particle states (out-states) |β. in|p′ . Proof: α. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . Proof: 0. thus β. . out . out| = eiϕ0 0. out (suppressing except momenta all other 1 m quantum numbers). (2) The one-particle state is invariant (conservation of energy and momentum. in = p. . out| and S † |α. in = S|α. e.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. x4 10. . p′ . x1 G(4. 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 ). ∂µ ∂ µ + M 2 φin (x) = 0. in|S = α.g ) (x1 . in|S = 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 . x3 . translation invariance). (10. in = p. out|α. in|S|p′ . out = |p′ . the vacuum expectation value of the ﬁeld φ(x) is zero. in = β. φout and the interpolating ﬁeld φ(x). in = p. out|p′ . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . out . in|S = 0.44) J=0 For the interacting theory. in| (choose ϕ0 = 0). out = p|p′ . implying the absence of tadpoles.g.45) x2 as the only surviving diagram (see exercises).Feynman diagrams for scattering amplitudes 92 i. x4 ) = c 1 x3 z (10. then (n) (n) Gsc = Gc for n ≤ 3. pn . out = δαβ ↔ SS † = 1.47) . in|SS † |α. For the free scalar theory. out|α. .

φin (y)] = iZ ∆(x − y). this can actually not be used as it would imply [φ(x). a causality condition that can be proven to imply the absence of interactions. a) = S. (10.g.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions.51) (10. β.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . The convergence therefore must be weakened to t→−∞ √ (10. out|φout = β. S is invariant under e Poincar´ transformations: U (Λ. The relation between S-matrix and in. (2π)3 2E (10. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10. the source term Jφ in the lagrangian density is treated in the interaction picture.Feynman diagrams for scattering amplitudes 93 while φ satisﬁes the inhomogeneous Klein-Gordon equation with the bare mass M0 (this is the mass appearing in the lagrangian L0 ). The relation between S and Z[J] To establish this relation. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. e. as it stands.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. 9.52) where Z is a constant. implies the strong (operator) convergence φ(x) → Zφin (x).e.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). The time evolution operator (see Eq. 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. z) φ(z) d4 y d4 z ∆A (x − y) K(y. i.54) . in|φin S → φin S = S φout (x).and out-ﬁelds is: φin (x) = S φout (x)S −1 . √ Although the above. 2 ∂µ ∂ µ + M0 φ(x) = J(x). Proof: β. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . (10. (10. a)SU −1 (Λ.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. φ(y)] = Z [φin (x). which in a later stage (renormalization) will become more important. 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. z) φ(z).

[A.e. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. φ− f (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]. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). S U [J]] = √ Z δ S U [J]. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. φ(y)] f (y) e e .58) In order to ﬁnd a solution to this equation note that. 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. which is equal to the expression e in which the creation operators are placed left of annihilation operators. φ(x).Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. δJ(z) (10. δJ(z) (10. C] are c-numbers. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. z) d4 y d4 z ∆(x − y)K(y.59) (10. just as we did for φ(x).55) δU [J] δJ(x) = t→∞ t→−∞ −→ −→ √ i φ(x) U [J] → i Z φout (x) U [J] √ i U [J] φ(x) → i Z U [J] φin (x). (10. eB ] = [A. we can. (10.63) . B] is a c-number). provided [A. e e ] = [A. y) δ δJ(y) : F [J]. B]eB . δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. z) . B + C]e e .62) where F [J] is some (arbitrary) functional. φ(x). one has [A. B] and [A.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. B] (for the case that [A. z) δU [J] .61) e φ+ f where the normal ordered expression : e φ f : is used. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x.57) i [φin (x).60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). express it in terms of advanced and retarded Green’s functions. t i Ut∞ [J] φ(x)U−∞ [J] (10. e = i. 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. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A.

. . |S|p. which is obtained considering only connected diagrams. . + pn ) G(n) (p1 .+ 1 √ Z ∗ d4 x′ . . . . pn ). . using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .. . .64) Therefore. pn ) = G (p1 .δ. (10.68) 10. (λ) v(p). d4 x . iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function.Feynman diagrams for scattering amplitudes 95 while for J = 0 one has U [0] = 1. (2π)3 2E n (10. Explicitly.. .2). ←− − → × G(n) (x′ . . u(p). which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. . . . . 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. . .e. . . Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . . . .65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. x. Usually we are interested in the part of the S-matrix describing the scattering.e. S =: exp 1 √ Z d4 x d4 y φin (x) K(x. . .e. . y). They start with the propagator (i∆F (k)) in momentum space. . u(p) and v (p) for fermions and ǫµ (p) for vector ﬁelds). pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . xn ). ¯ ¯ and the amputated or one-particle irreducible (1PI) Green’s functions n −i (n) Γ(n) (p1 . see exercise 10.) iKx . .. . iKx′ . i. . Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. c (n) (10. . Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. y) δ δJ(y) : Z[J] Z[0] . . . 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). ∆(pj ) j=1 (10. k = k2 i − M 2 + iǫ . . .66) where Gc is the connected Green’s function. p′ . . . = . . . 1 for scalar case.67) where ∆(p) is the Fourier transform of the (full) propagator ∆F (x).. J=0 (10. n (2π)4 δ 4 (p1 + . . i. fp (x) .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. . . . . . fp′ (x′ ) .

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. (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). which cancels the factor 1/2 in the quadratic piece.Feynman diagrams for scattering amplitudes 96 (in fact the inverse of the operator found in the quadratic term is for real scalar ﬁelds also multiplied by a factor 2. but overall momentum conservation at a vertex is understood. For the symmetry factor consider the examples (given in G. the factor 2 corresponds to the two-points Green’s function having two identical ends). For the interaction terms in the lagrangian. = −i g . ’t Hooft and M. If problems arise go back to the deﬁning expression for the generating function in 10. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically diﬀerent connected diagrams. to be precise iLI vertices in momentum space are introduced.27. vertices and external particle wave functions in that diagram. Diagrammar) in the case of the interaction terms g f (10. At these vertices each line can be assigned a momentum.66 and 10.27. for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar ﬁeld) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. Veltman.69) LI (φ) = − φ3 − φ4 . 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). Note that in calculating amputated Green’s functions external lines are neglected.

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

η]. = + . 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.. but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams. in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is (see next section for e-e-γ vertex).73) = i p / − M + iǫ = ji i(/ + M )ji p p2 − M 2 + iǫ The time ordered functions are obtained by functional derivatives from Z(η. η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ . which arises from the quadratic term in exp i d4 z LI . The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution. which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ ). (Rule 6) Each closed fermion loop gets a sign −1.71) = exp i where iSF is the Feynman propagator for fermions. 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 .. e. .72) (10. − g2 2 dz dz ′ δ2 δ2 δη(z) δη(z) δη(z ′ ) δη(z ′ ) and the quadratic term in Z0 [η. η).Feynman diagrams for scattering amplitudes 98 Dirac ﬁelds For fermions the generating functional is given by Z[η. (10.g.− i δη(z) i δη(z) d4 x d4 y η(x) iSF (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.. namely (Rule 5) Feynman diagrams which only diﬀer by exchanging identical fermions in initial or ﬁnal state have a relative minus sign.e.

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

s3 . where u ¯ L(m) µν = = 1 2 u(k ′ . s′ ) ¯ u s.76) 1 . If we are interested in the scattering process of an unpolarized initial state and we are not interested in the spins in the ﬁnal state we need |M |2 summed over spins in the ﬁnal state ( s3 . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ . s)¯(k.5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. s4 )(−ie)γ ν u(p. s4 )γµ u(p. ¯ q2 (10. s2 ). To lowest order (∝ α = e2 /4π) only one diagram contributes.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s1 ) u e2 u(p′ .s4 e4 (m) µν (M) L L .79) . t2 (10. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].Feynman diagrams for scattering amplitudes 100 10.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. s1 ) ¯ −igµν u(p′ . s′ )γµ u(k. s3 )γ µ u(k.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .s2 . s3 )(−ie)γ µ u(k. s2 )|2 ¯ q2 (10.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.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s)γν u(k ′ .

(b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. s2 (10. s2 )γµ u(k. . (c) Do the same for the connected Green function Gc . 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 10.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . momenta and invariants are e− e+ → µ− µ+ scattering k p s t = = (k + k ′ )2 = (p + p′ )2 (k − p)2 = (k ′ − p′ )2 (k − p′ )2 = (k ′ − p)2 k’ p’ u = The scattering amplitude squared (spins summed and averaged) is given by |M |2 = 1 4s e4 s2 |¯(p. This is known as crossing symmetry. s4 ) v (p′ . s2 ) ¯ ¯ 1 2s u(p. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s4 ) u e2 v (k ′ . The diagram. To lowest order (∝ α = e2 /4π) only one diagram contributes. (4) (4) . . the masses. s3 ) ¯ ¯ 1 .s2 3 . Similarly.s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . . s2 )γµ u(k.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 .s3 . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry.2. s1 ) u(k. Exercises Exercise 10. s3 )γ µ v(p′ .s4 = 1 2s × v (k ′ . s1 )γν v(k ′ .s2 . s1 )|2 ¯ s 1 . s3 )γ µ v(p′ . s4 )γ ν u(p.

. 1 2 (b) Calculate the quadratic pieces and interference terms. . Express the contributions in invariants s. xn ) ≡ (2π)4 δ 4 (p1 + . . t and u. . . .4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. . for instance.3 Show that the combinatorial factors found using the rules given in section 10.3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. . Note that the function G(n) (p1 . |M |2 = T11 + T22 − T12 − T21 . Exercise 10. . .2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. It is of the form −i M = A1 − A2 . . that the two-point Green’s function in momentum space has an incoming and an outgoing leg (as for the x-space Green’s function). xn ). pn ) only has n − 1 independent variables in it. .Feynman diagrams for scattering amplitudes 102 Exercise 10. + pn ). . their sum is zero. (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . which means overall momentum conservation in Green functions in momentum space. Show that i the amplitude is symmetric under the interchange of t ↔ u. . . . . . . e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . . pn ). Exercise 10. .38. . in the amplitude (Tij = A∗ Aj ).37 and 10. . + pn ) G(n) (p1 . . . or if we in general would consider all momenta as incoming. xn + a) = G(n) (x1 . . Pictorially this implies. . but the incoming and outgoing momenta are identical. hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . . xn ) ∝ (2π)4 δ 4 (p1 + .

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

12) pb = (mb . 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.15) Explicit calculation of the reduced two-body phase space element gives dR(s. 2mb (11. 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. (11. p1 . = 2 s λ(s.16) . m2 . which in the speciﬁc case a b also can be expressed as λ(s. 4π s 4π 8π s 4π (11. m2 ) = 4(pa · pb )2 − 4p2 p2 . m2 ) a b . 0). m2 . for instance.9) (11. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . 4s (11. p2 ) = CM = = 1 (2π)2 1 (2π)2 1 (2π)2 d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2 which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s. 2 s s − m2 + m2 a b √ .8) (11.11) (11. ptrf ).13) (11. p1 . Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . m2 . p2 ) = = where λ12 denotes λ(s. In that case trf pa = (Ea . m2 ) a b . 2mb λ(s. mb and s). m2 ) is a function symmetric in its three arguments. Also in this case one can express the energy and momentum in the invariants. m2 ). m2 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . a (11. m2 . 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = .10) The function λ(s.14) One can.

The variable s is always larger than a c b d the minimal value (ma + mb )2 .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.21) (11. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. tmax min λcd /4s. Consider the 1-particle inclusive case. usually is a 2-particle state. Instead of the scattering angle. Of course there are not only 2-particle channels. 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 .17) (11.27) with m2 = m2 − p2 = m2 + p2 + p2 . there is usually a preferred direction.19) (11. The initial state.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. in which one particle is detected. = (11. pz = mT sinh y. .20) → π0 + n → π+ + π− + n → .. 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 .22) (11. Examples appear in → π− + p (11. Writing E = mT cosh y. The various possibilities are referred to as diﬀerent reaction channels. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). At high energies. H1 + H2 → h + X. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. In inclusive measurements no particles are detected in the ﬁnal state. usually many particles are produced. pc . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . pd ) = 4π s 4π 8π s 4π dt dt √ √ . 11. py ). E . in exclusive measurements all particles are detected. 2s (11.. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. pT = (px . however.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt.18) (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees. Consider the process a + b → c + d.4).25) dR(s. (11. An often used set of invariants are the Mandelstam variables.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .

(11. uu = s) = Mab→cd (s. η = − ln(tan(θ/2)) = tanh−1 (cos θ). b ut = (pa − pd )2 = u. ut = u) = Mab→cd (s. 11. t. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . t. u). t. u = (p1 − p4 )2 = (p2 − p3 )2 . ¯ tu = (pa − pc )2 = t. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. tu = s. c bd Mad→c¯(su = u. (11.30) (11. ¯ tt = (pa − p¯)2 = s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). tt = s. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . 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.31) . ¯ b (11. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. This is known as crossing symmetry.28) which (only involving angles) is easier to determine. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. uu = (pa − p¯)2 = s. which means that rapidity distributions dN/dy maintain their shape. t-channel and u-channel processes respectively. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. u). t = (p1 − p3 )2 = (p2 − p4 )2 . Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t.2 Crossing symmetry In the previous chapter.

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

N and T · V are covariant. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . 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 . the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. . From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ).e. .34) i. N 1 . (11.35) This quantity is not covariant. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . (11. For the (proper) decay width one thus has 1 N Γ0 = . T · V ρa (11. σ= √ 2 λab T · V (11. (2π)4 δ 4 (Pi − Pf ) 2 = (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf ) d4 x ei(Pi −Pf )·x V. The 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 . . . 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 (11.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves).33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.T V.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. it is. as expected.38) (in which we can calculate −iMf i using Feynman diagrams).36) (11.Scattering theory 108 Although this at ﬁrst sight does not look covariant. dt (11.

43) dΩcm = π M (s. (11. 11. .47) M (s. for which the √ amplitudes squared have been calculated in the previous chapter.42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. see 11. 16π 2 s (11. implies for the scattering matrix M . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . 2 λab and for instance for two particles dσ = q ′ M (s.Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). . θ)|2 . (11. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . θ). q f ). p1 .e. for which one has dσ/dΩ = |f (E. δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . t) λab 4π dt. ˆ ˆ Inserted in the unitarity condition for M . 8π s 2π 8π s . θ) in quantum mechanics. . . . θ) = −8π s ℓ (in analogy with the expansion for f (E.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. (11. . .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s.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 . . . note the sign and cos θ = qi · qf ).41) (11.44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). (11. . The result is N = |Mf i |2 dR(s. respectively. pn ). . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. p1 . i. θcm ) √ q 8π s 2 2 (11. p1 . (11. q n ). . θ) or M (q i . pn ). q n ) q n qn √ Mni (q i .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .

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

s − M 2 + iM Γ (11. 0) ∝ e−i(E−iΓ/2)t . in which case. 11. however. η = 0. 0)|2 = e−Γt . For an unstable particle one expects that U (t. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. ∝ U (t. . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. For instance the tree-level calculation of Feynman diagrams at order e2 was real.e. . (11. the time-evolution operator. We note that unitarity is generally broken in a ﬁnite order calculation.55) (note that we have disregarded spin). The product of amplitudes is of order e4 . but still real. This is (somewhat sloppy!) seen by considering the (amputated) 1PI two-point vertex Γ(2) = −i ∆ as the amplitude −iM for scattering a particle into itself (forward!) through the decay channels as intermediate states. speciﬁcally one has for t > 0 that the Fourier transform is dk e 0 −ik0 t ∆(k) ∝ e−ik t dk (k 0 − Ek + iǫ)(k 0 + Ek − iǫ) 0 0 (t>0) ∝ e−iEk t . For the amplitude in a scattering process going through a resonance. 0). The prescription for the pole structure. such that |U (t. 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. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11.56) (again disregarding spin). it is straightforward to write down the partial wave amplitude.58) . . which is given by a sum of the partial widths into the diﬀerent channels. .e. (q Mℓ )ij (s) = −M Γi Γj .57) This shows that Γ is the width of the resonance. i. To check unitarity we need the amplitude at order e4 . i. σel = σinel . The quantity Γ is precisely the width for unstable particles. This is what we will do to discuss decay widths. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . 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 .

At resonance the cross section σT (π + p) is about 210 mb. This characteristic shape of a resonance is called the Breit-Wigner shape. dt s(s − 4m2 ) with f (t. Exercises Exercise 11.e. a fast change in the phase shift for a narrow resonance. t)} . Its mass is M = 1232 MeV. the full width is Γ = 88 keV. Limiting ourselves to a resonance in one channel. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . which is attributed to neutrinos.49 GeV. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. 2 q (s − M 2 )2 + M 2 Γ2 (11. All these decays can be seen and leave an ’invisible’ width of 498 MeV. The cross section at resonance is about 30 nb. where furthermore σinel = 0. Γ = 2.4) can be written as 4πα2 dσ = {f (t. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . Three neutrinos explain the resonance shape.1 Show that the cross section for electron-electron scattering (exercise 10. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. • The J/ψ resonance in e+ e− scattering. u) = g(t. Its mass is M = 3096. i. the partial width into e+ e− is Γee = 5. Knowing that each neutrino contributes about 160 MeV (see next chapter). t) + g(u. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. u) + g(t. • The Z 0 resonance in e+ e− scattering with M = 91. This is a narrow resonance discovered in 1974. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. Note that because of the presence of a background the phase shift at resonance may actually be shifted.88 MeV.Scattering theory 112 (Prove this using the unitarity condition for partial waves).60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. u) + f (u.26 keV.2 GeV. 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). M2 − s (11. its width Γ = 120 MeV.59) reaching the unitarity limit for s = M 2 . The half-width of the resonance is M Γ.

Calculate now fπ .8. Exercise 11. Exercise 11. Compare the results with the experimental π − lifetime τ = 2. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. Thus.2 Calculate the ﬂux factor.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. writing N (s) M1 (s) = 2 + iM Γ .Scattering theory 113 Exercise 11.4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ . a quantity which we will ¯ ¯ encounter in the next chapter (section 12.000 123 Γ. ρcm ρcm a b 1 cm .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. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . 4. cm |va − vb | in the center of mass system for two colliding particles a and b.999 877 Γ and Γ(π − → e− νe ) = 0.2). As discussed for unstable particles.

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

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

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

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

inf

(12.3)

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

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

The standard model

115

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

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

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

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

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

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

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

**A geometric picture of gauge theories
**

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

1 The

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

λ7 =

i

−i

1 1 λ8 = √ 3

−2

= . The relation in Eq. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. but expresses them with respect to diﬀerent bases. i. ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). should be independent of such transformations.e.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x).14) (12. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). (12.16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). (12. -dx -dy µ µ dyµ µ dx For a vector. (12. 12. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed.The standard model the basis for x G and x+dx 117 G. ds which is solved by dψ(s) ds = dxµ . However.16. which connects two identical vectors. 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. In principle such a phase in a given point is not observable.e. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . (12. i.19) This gives rise to a (path dependent) phase in each point. ds s dxµ ds′ P exp ig A (s′ ) ψ(0).15) ψ(x).

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

ϕc = 0|φ |0 = (12. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. ϕ2 and η. i. The situation now is the following. They are known as the Weyl mode or the Goldstone mode: Weyl mode. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . In this situation one speaks of spontaneous symmetry breaking. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . (12. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . 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. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized.23) F +η the latter only forming a minimum for m2 < 0. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. It is only invariant under rotations around the 3-axis. The classical groundstate appears degenerate. In this case the spectrum is described . . In the case of degeneracy. has less symmetry.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 .25) 2 2 2 Therefore there are 2 massless scalar particles.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. .e. while the lagrangian including the nonzero groundstate expectation value chosen by nature. say 0 0 . In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . while the quadratic terms must correspond with non-negative masses. The result is 1 1 1 1 (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 − m2 (ϕ2 + ϕ2 ) L = 1 2 2 2 2 2 1 2 1 − m (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . etc. for the vacuum Qa |0 = 0.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + .) terms constitute interaction terms. therefore a choice must be made. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. Any constant ﬁeld with ’length’ F is a possible groundstate. . (12. (12. The η massless particles are called Goldstone bosons.

Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). 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. Neglecting at this point the local color symmetry. . i. H] = 0.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x).The standard model 120 by degenerate representations of the symmetry group. ψu mu md ψ = ψd . [Qa . irrespective of the fact if they annihilate the vacuum. .26) for all the states labeled by a corresponding to ’broken’ generators. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . Including a sum over the diﬀerent ﬂavors (up. (12.. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. T ] ψ. down. V µ = ψγ µ T ψ.31) . 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 conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).32) (12. denoted |π a (k) . As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. it is easy to see that one gets ∂µ V µ = i ψ [M. = ψ(i/ − M )ψ.e. π (12. ∂ ∂ (12. etc. called ﬂavors. 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. T = τ /2.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. a ∂ µ Jµ = 0. (12. i. A bit more formal.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). i.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. (12. ψ −→ ei α·T ψ. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. if the generators Qa generate a symmetry. a and a′ are degenerate states. The generators Qa (in that case the rotation operators Lz .e. a massless Goldstone boson for each ’broken’ generator. Using the Dirac equation. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . Taking the derivative. and one must have mπa = 0. . Goldstone mode. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. there must be a nonzero expectation value 0|Jµ (x)|π (k) . . M = . This means that they are operators that create states from the vacuum. strange.) one can write L = f ψ f (i/ − Mf )ψf .e.

Note that these transformations also work on the spinor indices. much smaller than any of the other mesons or baryons. In the real world. parity minus). e.e. etc. while the groundstate is only invariant under isospin rotations (R = L). but the fact that the symmetry is not perfect and the right hand side of Eq.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. it is easy to see that one gets ∂µ Aµ = i ψ {M. There exists another set of symmetry transformations.35 is nonzero. is what happens in the real world. ψ −→ ei α·T γ5 ψ.34) Using the Dirac equation.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . M = m · 1.e. Therefore the world of up and down quarks describing pions. with almost degenerate masses (Mp = 938. (12. etc. including in Dirac space a γ5 matrix.e. socalled axial vector transformations. How can we see this. 12. This symmetry. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. which is approximately true for the up and down quarks.27. i. is by nature not realized in the Weyl mode. ψR/L = 1 (1 ± γ5 )ψ.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . ∂ ∂ (12.36) 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. The chiral ﬁelds ψR and ψL are transformed into each other under parity. both are very small. This combined symmetry is what one calls chiral symmetry. parity would not play a role in the world. Aµ = ψγ µ T γ5 ψ. T ] = 0. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. if in ﬂavor space M is proportional to the unit matrix. fπ = 0 remains. i.e. From group theory (Schur’s theorem) one knows that the latter can only be true. for which the ﬁeld (according to the discussion above) has the same behavior under parity. however.6 MeV/c2 ). behaves as a pseudoscalar particle (spin zero. but also a triplet (isospin 1) of pions. generated by TR/L = 1 (1 ± γ5 )T . This situation. stated equivalently. a doublet (isospin 1/2) of nucleons. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . as the quantity ∂µ Aµ (x). or. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. 12.3 MeV/c2 and Mn = 939.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. (12.g. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. proton and neutron. the lagrangian density is invariant under left (L) and right (R) rotations independently. T } γ5 ψ. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. From the number of broken generators it is clear that one expects three massless Goldstone bosons. (leaving out the ﬂavor structure) the same as ψγ5 ψ. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. where the quark masses are not completely zero. T γ5 = τ γ5 /2. which is equivalent 2 to the V-A symmetry. i. chiral symmetry is not perfect. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . I. (12. . This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry.

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

R −→ e−i β R. however. down. the leptons (elektrons. etc.51) . 2 (12. muons. νµ and ντ ) and their antiparticles. The procedure. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 .48) (12. τ ). neutrinos. its corresponding neutrino (νe . i. 2 2 = ∂µ R + i g ′ Bµ R. ψR/L = 2 (1 ± γ5 )ψ decouple. ∂ ∂ ∂ (12. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.The standard model 123 12.50) (12.49) 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 /2.4 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. (12. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.e. = ∂µ L − (12. etc. 1 As they are massless. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. the quarks (up.) and the gauge bosons responsible for the strong. left. (12. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. muon µ− or tau τ − ). R. This is written as 3 Q = TW + YW .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. explicitly L −→ e−i β/2 L.). U(1)Y U(1)Y (12. 0 1 +1/2 L νe 3 − = with TW = and TW = L= −1/2 eL L 2 R = R − = eR 3 with TW = 0 and TW = 0.44) which has an SU (2)W symmetry under transformations eiα·TW . Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . electromagnetic and weak forces.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist.and righthanded particles. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. while the lefthanded particles form a doublet.47) and YW is considered as an operator that generates a U (1)Y symmetry. that gives a good description of the physical world.45) (12.

the U (1)Q symmetry (generated by the charge operator). 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. h) (12.56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms. (12. L where L (h1) (h) =L (h1) +L (h2) . 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. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. .57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .59) . 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. . Putting this in leads to L (f ) =L (f 1) +L (f 2) . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. 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 + . leading to the parametrization 1 0 φ(x) = √ (12.58) (12. 4 4 In order to break the symmetry to the symmetry of the physical world. −V (φ) L and (h2) = −Ge (LφR + Rφ L). .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 .55) 2 2 The Higgs potential V (φ) is choosen such that it gives rise to spontaneous symmetry breaking with ϕ† ϕ = −m2 /2λ ≡ v 2 /2. + 8 (12.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . Using local gauge invariance θi for i = 1. this leads to the lagrangian L (h1) . (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1.

67) The coupling of electrons or muons to their respective neutrinos. we can read oﬀ e = g sin θW = g ′ cos θW .61) (12. .60) (12.68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ .63) (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 . .69) . 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2 (12. (12. 4 2 MW g2 v2 = . g ′ /g = tan θW . for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + . (12. 2 Zµ (12.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 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 . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.65) The weak mixing angle is related to the ratio of coupling constants. and correspond to three massive particle ﬁelds (W ± and Z 0 ) and one massless ﬁeld (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 .64) (12. 2θ 2θ 4 cos W cos W 0.66) From the coupling to the photon.

231 2. i. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. 80.40 GeV.g.75) The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with 3 TW 1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 . (12. Comparing this with the total width into (invisible!) channels. 2 2 2 (12. (12. (12. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.80) First. (12. from the MW mass is about 250 GeV.166 4 × 10 0. 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 ).70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.e.71) GeV −2 = = = = . The coupling to the Higgs particle is weak as the value for v calculated e.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. . giving rise to a running coupling constant after renormalization of the ﬁeld theory.74) (12.19 GeV. v (12. one has the relation g2 e2 1 GF √ = . g ′ and v determine a number of experimentally measurable quantities.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.72) (12. Finally we want to say something about the weak properties of the quarks.76) with CV CA 3 = TW − 2 sin2 θW Q.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. i. 3 = TW . g2 MZ 2 2 (CV + CA ). such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. −5 (12. muon or tau. 91. Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three.77) (12.5 MeV (exp. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . With the choosen vacuum expectation value for the Higgs ﬁeld. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).e. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. me /v is extremely small.73) (12. 1.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.

t c u ′ . this requires particular YW -TW assignments to get the charges right. their antiparticles and the electroweak gauge bosons as appearing in each family. 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. The . Comparing neutron decay and Λ decay one can get an estimate of the mixing parameter Vus in the socalled Cabibbo-Kobayashi-Maskawa mixing matrix.3. leptons. suggesting an underlying larger unifying symmetry group. 12. Decay of B-mesons containing b-quarks allow estimate of Vub . ′ ′ 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. etc.81) n −→ pe− νe ⇐⇒ d −→ ue− νe . We will not discuss this any further in this chapter. which can account for the (observed) CP violation of the weak interactions.The standard model 127 YW eL uR νR +1 eR dL dL _ uL W 0 W −1/2 B0 +1/2 W + I3 W uR dR eL −1 dR Q = +1 νL Q=0 uL 3 Figure 12. The pattern is actually intriguing. but with diﬀerent strength. Λ −→ pe− νe ¯ ⇐⇒ s −→ ue− νe . The electric charge Q is then 3 ﬁxed.3: Appropriate YW and TW assignments of quarks. Q = TW + YW /2 and constant along speciﬁc diagonals as indicated in the ﬁgure. i. As shown in Fig. 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. In principle one complex phase is allowed in the most general form of the CKM matrix. This is only true if the mixing matrix is at least three-dimensional. for which SU (5) or SO(10) are actually nice candidates. ¯ 3 This mixing allows all quarks with TW = −1/2 to decay into an up quark.e.

u u where we include now the three lefthanded quark doublets in QL .84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. The Higgs ﬁeld is still limited to one complex doublet.82) space.g.e. ms and mb ).82 and ρ ≈ 0. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. 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 . e. .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . (12.83) where Vu and Vd are unitary matrices. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. † Λu Λ† = Vu G2 Vu .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0.227. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. UR = uR cR tR . for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .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. i. the weak mass eigenstates are L (h2. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . respectively). 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. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. A ≈ 0.86) the unitary CKM-matrix introduced above in an ad hoc way. (12. 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 . each of these containing the three families. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .22 and η ≈ 0.34.

† † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . we ﬁnd massless neutrinos. since the weak current is the only place where they show up.92) .89) with three righthanded neutrinos added to the previous case. mµ and mτ . (12.e. decoupling from all known interactions. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. EL = Ve† EL . 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 . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ).90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . (12.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. i. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. The mass ﬁelds ER † mass † mass = We ER . and obtain L (h2. 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 . e N L= L EL (12. (12. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . EL = Ve EL . UPMNS a parametrization that in principle also could have been used for quarks.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . there is no family mixing for massless neutrinos. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . 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. (12.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.ℓ) = −LφΛe ER − ER Λ† φ† L. In this case. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . As before. one can write Λe = Ve Ge We and we ﬁnd † L (h2.93) 1 = 0 0 0 c23 −s23 0 s23 c23 c13 0 −s13 eiδ 0 s13 eiδ 1 0 0 c13 c12 −s 12 0 s12 c12 0 0 0 1 .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .5×10−3 eV2 .

96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . For the neutrino sector. couple the right. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors.ν =− containing three (CP-violating) phases (α1 . We use here the primes starting with the weak eigenstates. 0 0 1 L mass. Absorption of phases in the states is not possible for Majorana neutrinos. it is easy to see R that this incorporates the Dirac case by considering the lefthanded part of Υ′ and the righthanded 1 part of Υ′ as a Dirac spinor ψ. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry.95) 2 In order to have more than a completely decoupled sector. as above. − MD MR nR 2 (12.ν = − MR NR NR + MR NR NR . however.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. The mass matrix can be written as ML |MD | eiφ M = (12. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. is to add in Eq. (12.98) (12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. Actually. This is called the see-saw mechanism (outlined below). 2 although this option is not attractive as it violates the electroweak symmetry.97) As the most general mass term in the lagrangian density we have L mass = − = which for MD = 0 is a pure Majorana lagrangian and for ML = MR = 0 and real MD represents the Dirac case. 1 ∗ c c (12. hence the mixing matrix becomes iα /2 0 0 e 1 (12.and righthanded species then just form a Dirac fermion. Thus 2 Υ ′ = nc + nL . a unitary matrix relating them to the weak eigenstates. For the leptons. 1 c ∗ c L mass. For these light Majorana neutrinos one has. α2 and δ). Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). coupled by a Dirac mass term.94) M L NL NL + M L NL NL . (12.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. one must for the neutrinos as well as charged leptons. the massless and massive Majorana neutrinos. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.c. Thus (disregarding family structure) one has two Majorana neutrinos.99) . with for the righthanded sector a mass term MR .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. one being massive. are equivalent to two decoupled Majorana neutrinos (see below). 2 R ψ = nR + nL . 1 L Υ ′ = nR + nc . the left. 12.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section.

2 with real masses Mi . H] = 0. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. a and a′ are degenerate states. [Qa .105) for i = 1. (12.The standard model 131 taking ML and MR real and non-negative.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. 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.1 Consider the case of the Weyl mode for symmetries. 2 U (M M † ) U † = M0 . Exercise 12.2 Derive for the vector and axial vector currents.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. i ψ {M. Exercises Exercise 12.101) Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2 . Prove that if the generators Qa generate a symmetry. (12. Υ2R nR (12.e. T } γ5 ψ.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . i. T ] ψ. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . i. related via c Υ1R = U nL . L Υ2L nR Υ1 and Υ2 . . (12. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . 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).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)) . 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′ .

Exercise 12. Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 . mb ≈ 5 GeV and MW ≈ 80 GeV. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 .6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . (b) Calculate the width to electron-positron pair. Mf M2 and ghf f = . Exercise 12. 2 2 The masses are mt ≈ 175 GeV. e+ e− h. hhh.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . . 3 TW . ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld. The mass of the Z 0 is MZ = 91 GeV. at least up to a high (which?) order in λ. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . Γ(Z 0 → νe νe ).) are proportional to the mass squared for bosons or mass for fermions of the particles. and the width to a pair of neutrino’s. to be precise one has for bosons (b) or fermions (f). Exercise 12.231. Exercise 12. etc.5 Show that the couplings to the Higgs (for W + W − h.7 In this exercise two limits are investigated for the two-Majorana case. ZZh. Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. Γ(Z 0 → e+ e− ). the weak mixing angle is given by ¯ sin2 θW = 0.

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