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

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

November 2011 (version 6.04)

Contents

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

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

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

1

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

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

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

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

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

2 2. 6. 4. Quantum Field Theory. 3. I: Foundations.2 2. Peskin & Schroeder and Srednicki. Zuber.1 3. 1996. Veltman. Corresponding chapters in books of Ryder. 1980.1 2 2 3.2 3. McGraw-Hill.3 References As most directly related books to these notes. Schroeder.3 2.2 3.2 3. 2. but are rather compact. 2.4 3. Quantum Field Theory.4 2. 5.1 2. Vol.5 4.3 3. I refer to the book of Srednicki [1] and Ryder [2].2 1. 1995. 3. M. Quantum Field Theory. Weinberg.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. S.2 36 22 .2 3.3 3.5 2. Ryder. M.5 2.E.7 2.4 3.1 3.1 6. Vol.4 3.-B. 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). Addison-Wesly.3 6.3 3. The quantum theory of ﬁelds.2 4. An introduction to Quantum Field Theory.3 2.1 4. 1994.3 2. Cambridge University Press.H. Cambridge University Press.5 5.3 2.4 4.2 2.1 5. Cambridge University Press. Peskin and D. Cambridge University Press.V. 1985. Itzykson and J. C. 1995.1.2 6.3 2. L.4 2.3 3.1 2.3.3 4. M. The notes contain all essential information.2.2 3. 2007. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. 1. Diagrammatica.2 6. II: Modern Applications. Srednicki. These notes Section 1.3 3.1 1. 3. Cambridge University Press.4 6.8 3.

2 8.3 8. 4.4 4.4 10.1 12.5 .4.4 12.3 12.1 8.7.1 10.8 5 4.4 These notes Section 7.1. 20.2 9.3 9.7 5. 4.6 9. 6.2 10.3 11.1 5.1 9.4 2.1 9.3 10.1. 9.6 3.3 6 8 9 Srednicki 3 3 3 37 5.4 8.5 6. 9.8 6.2 12. 2.2.5 4.5 7. 6.2 4.1 20.6 6.3 2.1 20.2 11.3.3.3 6.2. 4.4 9.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.1 11.6 3.6 4.1 20.6 10.5 6 7 6 6.5 2. 2. 6.2.5.2 8. 8.5 4.1 7.1.4 12. 6. 4.5 9.3.3 3.5 7.2 7. 6.6 8.6 3.4 7.5 11.4.2 5.4 3.2.3 7.7 4.3 7.3 20.10 8.4.5 9.3 8.

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

In quantum mechanics. Indeed. S ≫ .4) where δij is the Kronecker δ function.582 119 15 (56) × 10−22 MeV s. One can talk about states |r and the wave function ψ(r) = r||ψ .5) One can also choose a representation in which the momenta of the particles are the dynamical variables. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). (1. In quantum mechanics a special role is played by Planck’s constant h.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. ≡ h/2π = 1. (1. ˜ ψ(p) = i d3 r exp − p · r ψ(r).1) In the limit that the action S is much larger than . (1. (1. They satisfy the well-known (canonical) operator commutation relations [ri .3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. the position can in principle be determined at any time with any accuracy. the action of the momentum operator in the coordinate representation is not as simple as the position operator. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-deﬁned concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. In special relativity a special role is played by the velocity of light c. (1.2) In the limit that v ≪ c one reaches the non-relativistic domain. 1 (1.054 571 68 (18) × 10−34 J s = 6.Chapter 1 Introduction 1. quantum eﬀects do not play a role anymore and one is in the classical domain. c = 299 792 458 m s−1 . ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. pj ] = i δij . Corresponding position and momentum operators do not commute. It is given by pop ψ(r) = −i ∇ψ(r).6) . being eigenvalues of the position operators. usually given divided by 2π.

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.

(1.. 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 . ˆ ˆ (1. independent of a coordinate system. δim δjm δkm δin δjn δkn . such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . The (invariant) lengths often have special names. 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. x2 .22) (1. where the coordinate t = x0 is referred to as the time component. • ds2 = 0 (lightlike or null intervals).3).24) The quantity gµν ≡ nµ · nν is the metric tensor. .21) (1. The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations. When 3-dimensional space is considered as part of Minkowski space. x1 . The Poincar´ transformations include e e Lorentz transformations and translations. and are given by (x0 . (1. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations.νn can be conν structed. One can distinguish: • ds2 > 0 (timelike intervals). x2 .. x3 ) = (t. Because of the diﬀerent signs occurring in gµν . xi are the three space components. x3 ). The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .2. We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. one must distinguish tensors with upper or lower indices and one can have mixed tensors. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . These transformations do change the components of a vector. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . the invariant length or distance (squared) is not positive deﬁnite. The lower indices are constructed in the following way. In special relativity we start with a four-dimensional real vector space E(1.1.3) with basis nµ (µ = ˆ 0. So we could have used all upper indices in the above equations. −x). we will use upper indices for the three-vectors. . The length (squared) of a vector is obtained from the scalar ˆ product. One has x2 = xµ xµ = t2 − x2 . denoted as V ′µ = Λµν V ν . it is convenient to distinguish lower indices from upper indices. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. are .25) Within Minkowski space the real. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. in which case zi = ǫijk xj yk . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero).23) δjm δkn − δjn δkm . xµ = gµν xν . • ds2 < 0 (spacelike intervals). x) = (x0 . however.. Relations relating tensor expressions.. Many other four vectors and tensors transforming like T ′ µ1 . x1 .µn = Λµ1ν1 . Vectors are denoted x = xµ nµ . The real. the points lie on the lightcone and they can be connected by a light signal. No diﬀerence is made between upper or lower.Introduction 5 that can be used in the cross product of two vectors z = x × y. In Minkowski space. Unlike in Euclidean space. Λµn n T ν1 . 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 .26) (1.

34) (1. p) = (E. = ∂ .∇ . (1.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). 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. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. . p). The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . starting from ǫ0123 = 1. (1.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . a∞ = 4πǫ0 2 /me e2 . For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .Introduction 6 called covariant. = −24. ∂2 . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . ′ ′ . (1. (1. ∂1 . . but more important has the same eﬀect on lefthandside and righthandside. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. The length squared of ∂ is the d’Alembertian operator. g0 = g1 = g2 = g3 = 1.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.32) . . ∂3 ) = ∂ ∂ ∂ ∂ . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . explicitly written as (p0 .33) (1. ∂t (1. not aﬀected by Lorentz transformations.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. where m is called the mass of the system. 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 .28) ν ν Note that gµ with one upper and lower index. Each of these permutations leads to a minus sign. constructed via the metric tensor itself.27) It is easy to convince oneself of the nature of the indices in the above equation. (1.36) Exercises Exercise 1. because one has (1. for instance. Note that in this equation one has on left. It is an invariant tensor. deﬁned by ∂2 − ∇2 .29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .

]. Also construct and calculate the Planck length Lpl . Exercise 1. ρ. e. 1.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). a1 . n1 . 1. (e) Note that /m = − c and use this to calculate in a simple way (avoiding putting in the value λ of e) the Bohr magneton µe = e /2me and the nuclear magneton µp = e /2mp in electronvolts per Tesla (eV/T). 1. 0) ˆ n2 = (0. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . [Note: what is the MKS unit for V/T?] Exercise 1. a3 ) are the expansion coeﬃcients using the basis vectors n0 . . 0. a+ . Also for the coordinates [a− . ˆ ˆ ˆ ˆ These are n1 = (0. n+ . c. (b) The coordinates (a0 . ǫ0 . a1 . g−− . re = e2 /4πǫ0 me c2 to the Compton wavelength. (d) Use the value of the gravitational constant GN / c5 = 6.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . Compare it with the proton mass and use Eq. 3} the index α can only be equal to one of the indices in ǫµνρσ . σ} is a permutation of {0. a+ . . which is the Compton wavelength for the Planck mass. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . n1 . n2 . n3 . ν. a+ . One does not need to know . 0) ˆ n3 = (0. ˆ ˆ . 1. 0) ˆ n0 = (1. 0. me . a2 ] we can ﬁnd basis vectors n− . 2. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. This demonstrates how a careful use of units can save a lot of work. a1 . Exercise 1.2 Prove the identity A × (B × C) = (A · C) B . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. 0. by a simple few-line reasoning [For instance: If {µ.3. α and me c2 = 0.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . a2 ] or [a− . How are a+ and a− related to a+ and a− .511 MeV. g+− and g−+ . . but only appropriate combinations.13 to give its actual value in kg. n2 . 0.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). 1) ˆ . 0.(A · B) C using the properties of the tensor ǫijk given in section 1. a2 . 0.

7) k 0 = ± k2 + M 2 = ±Ek .2.1 The Klein-Gordon equation In this chapter. ∂t p −→ pop = −i∇. E = p2 . Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. i ∂ ∇2 ψ(r. ∂t 2M (2.3) Equations 2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. written as pµ −→ i∂µ is covariant (the same in every frame of reference). t) = − ψ(r. (2.3 are not covariant. 8 . (2. ∂t2 (2. 2M (2. t) = ∂2 − ∇2 + M 2 φ(r. t). 2 + M 2 φ(r. (2. t) = 0.4) where M is the particle mass. But the replacement 2. t) = exp(−i k 0 t + i k · r).1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . Although it is straightforward to ﬁnd the solutions of this equation.5) with (k 0 )2 = k2 + M 2 . namely plane waves characterized by a wave number k.1 and 2. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite. φk (r.2) Acting on the wave function one ﬁnds for a free particle. one obtains (2. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. p2 = pµ pµ = E 2 − p2 = M 2 . p).Chapter 2 Relativistic wave equations 2.

which is what Lorentz transformations do. we want the most general solution.13) ˜ with (in principle complex) coeﬃcients φ(k). The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. The KG equation. The appropriate current corresponding to the KG equation (see Excercise 2. (2π)3 2Ek (2. j) = ↔ i φ∗ ∂0 φ. ↔ ↔ (2. (2.11) Therefore.9) They satisfy the continuity equation. As we will see later both problems are related and have to do with the existence of particles and antiparticles. −i φ∗ ∇ φ . k) + ei k·x φ(−Ek . just as the dependence on time was in ordinary quantum mechanics. k) ≡ a(k) and φ(−Ek . is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. relativistically the density is not a scalar quantity. 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. ∂µ j µ = 0.3). however. This continuity equation can be written down o covariantly using the components (ρ. −k) . This operator has all possible solutions in it multiplied with creation (and annihilation) operators.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds.e. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. And. Then. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. there are solutions with negative energy. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. there are no longer fundamental objections to mix up space and time. 2. for which we need the interpretation of φ itself as an operator. 2.2) is j µ = i φ∗ ∂ µ φ or (ρ. j) of the four-current j.5). rather than as a wave function.8) (2. (2. but rather the zero component of a four vector. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. having the KG equation as a space-time symmetric (classical) equation.15) .Relativistic wave equations 9 i. (2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x).10) ∂t which follows directly from the Schr¨dinger equation. 2M 2M (2.14) ˜ ˜ Introducing φ(Ek . ∂ρ = −∇ · j. This leads to the existence of negative densities. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek .

which implies (xµ ) = (t. after restricting the energies to be positive).20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k).6 We will explicitly discuss the example of a discrete symmetry. φP (x) ≡ φ(˜) (P for parity). Since a · b = a · ˜ it is easy to see that ˜ b.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . 2. Another symmetry is found by complex conjugating the KG equation.3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. (2π)3 2Ek (2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). x ∂µ ∂ µ + M 2 φ(˜) = 0.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. Performing some Lorentz transformation x → x′ = Λx. 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) .22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). (2π)3 2Ek (2. (2. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. i. This shows how parity transforms k-modes into −k modes.17) (2. changing the sign of the spatial coordinates. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.16) The consequence of this is discussed in Exercise 2.Relativistic wave equations 10 In Eqs 2.23) . −x) ≡ (˜µ ). They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). x) → (t. (2.e. 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) .15 one has elimated k 0 and in both equations k · x = Ek t − k · x. (2π)3 2Ek (2.18) (2.14 and 2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). for which we consider space inversion. (2. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.

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

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). . t) = (−i α · ∇ + m β) ψ(r. {αi . and β 2 = I. i. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). a{µ bν} ≡ aµ bν + aν bµ . We will encounter these quantities later as energy and momentum. ∂ ψ(r. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. αj } = 2 δ ij I. t). (c) Similarly. show that for two ﬁelds φ1 (with coeﬃcients a1 and b∗ ) 1 and φ2 (with coeﬃcients a2 and b∗ ) one has 2 d3 x φ∗ (x) φ2 (x) 1 = d3 k 2 (2π)3 4Ek a∗ (k)a2 (k) + b1 (k)b∗ (k) 1 2 + a∗ (k)b∗ (−k) e2iEk t + b1 (k)a2 (−k) e−2iEk t . B] = AB − BA and the anticommutator as {A. From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. 1 2 (b) Express now the full charge QV (t) (exercise 2. d3 x (∂ {0 φ)∗ (∂ i} φ). B} = AB + BA.5 (a) As an introduction to parts (b) and (c). Note that a{µ bν} indicates symmetrization.e.7 To solve the problem with positive and negative energies and get a positive deﬁnite density. i (a) Show that relativistic invariance. also using the result in (a). Exercise 2. Exercise 2. in terms of the a(k) and b(k). 2.Relativistic wave equations 12 Exercise 2. making sure that each component of ψ satisﬁes the KleinGordon equation (in terms of diﬀerential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi .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).1) for a complex scalar ﬁeld current (exercise 2.15. j 0 = ψ†ψ and j = ψ † αψ. β} = 0.2) 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 . ∂t where ψ is a wave function and the nature of α and β is left open for the moment. 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).

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

It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . and thus (σ · n)2 = 1.7) = a0 0 −1 i 0 1 0 0 1 • [x.and y-axes 0 0 0 0 0 0 0 −i . [y. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ.1: the parameter spaces of SO(3) (left) and SU (2) (right). These generators form a threedimensional Lie algebra SO(3). with real a’s and 3 (ai )2 = 1. Next. They satisfy the commutation relations [Li . x]] + [z. y] ∈ A. z ) = lim N →∞ that are generated by i 0 . (3. is a 3-dimensional Lie-group. z]] + [y. 0 (3. for rotations around z for instance. ] that satisﬁes • ∀ x. Lj ] = i ǫijk Lk . y]] = 0 (Jacobi identity). y] = −[y. • [x. One way of viewing the parameter space. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. Ly and Lz . Locally this is a 3-dimensional Euclidean space and SU (2). x]). x] = 0 (thus [x. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . In the same way we can consider rotations around the x. [z.z N N or (Lk )ij = −i ǫijk .6) Summarizing.2 π) (-n.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). ˆ R(ϕ. y ∈ A ⇒ [x. namely that there exists a bilinear product [. π) (n. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . ˆ therefore.5) Rotations in general do not commute.4) R . π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. which reﬂects itself in the noncommutation of the generators. (3. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices.Groups and their representations 14 (n. . 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. [x. 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† ).

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

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

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

23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index.22) (Note that xT is a row vector). it is convenient to use a vector notation for the points in Minkowski space. (det Λ = +1 is called proper.25) (3. Writing x as a column vector and the metric tensor in matrix form. e To derive some of the properties of the Lorentz group. G = 10 (3. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ . (3. From this property. 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. In this section we consider the Poincar´ group.21) g 20 g21 g22 g23 0 0 −1 0 g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . We considered the generators and looked for representations in ﬁnite dimensional spaces.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. det Λ = −1 is called improper). (ii) |Λ00 | ≥ 1. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.26) (3.g. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. consisting of the Lorentz group and translations. σ/2 in a two-dimensional (spin 1/2) case.27) (3. e.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone .Groups and their representations 18 3. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.

J = (J 1 . since the generators K do not form a closed algebra. inﬁnitesimally given by ˆ ΛR (ϕ. In L↑ . z ) = exp(i ϕ J 3 ). 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η.Groups and their representations = i=1 (Λ0 )2 . one distinguishes rotations 0 + and boosts. K j ] = −i ǫijk J k . Using these explicit transformations. we have found the generators of rotations. [J i . z ) ≈ I + i ϕ J 3 . K 3 ). Returning to the global group. inﬁnitesimally given by ΛB (η. 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).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↑ . J 3 ). K 2 . (3. K j ] = i ǫijk K k . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. 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. (3. J 2 . J j ] = i ǫijk J k . and those of the boosts. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. it is easy to ﬁnd the following typical examples from each of the four classes. Thus ˆ 0′ 0 0 0 V 0 V 1 0 0 0 0 1′ 1 V 0 cos ϕ sin ϕ 0 V 0 0 −i 0 2′ = 2 −→ J3 = V 0 − sin ϕ cos ϕ 0 V 0 i 0 0 . 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. z ) = exp(−i ηK 3 ). This is the origin of the Thomas precession. 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 . which satisfy the commutation relations (check!) [J i . βγ = sinh η. The commutator of two boosts in diﬀerent directions (e. K = (K 1 . since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. z ) ≈ ˆ ˆ I − i η K 3 .30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3.28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞.32) 0 − 0 −1 0 0 0 0 −1 . Rotations around the z-axis are given by ΛR (ϕ.g. i proof: use ΛT GΛ = G and ΛGΛT = G. [K i .

P+ etc. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ .38) Explicitly. ǫ). (3. We therefore ﬁnd (again) that there are six generators for the Lorentz group. In order to ﬁnd the commutation relations including the translation generators. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. We will just require Poincar´ invariance for the generators themselves. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts).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) (3.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .39) (e. we continue using covariance arguments. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. The extension to the Poincar´ group is e ↑ straightforward.36) (3. K = M ).g. 3. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. Of the four parts only L↑ forms a group. inﬁnitesimally approximated by (I + ω. a). One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . [M µν . i i0 i i0 1 ijk M jk 2ǫ (3. Also this group can be divided into four parts. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . There are six generators.g. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . 3 30 with η = ω and K = M (e.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . This is a + normal subgroup and the factor group is the Vierer group. (3. Summarizing. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). inf (3.33) These four transformations form the Vierer group (group of Klein). we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ.

P µ |p. P j ] = i ǫijk P k . . that commute e among themselves. (3. P j ] = [P i . . (3.47) (3. H] = 0. These form a group called the little group associated with that four vector. 2 2 (3. [J i . .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . one obtains 2 [P i .41) 2 At this point. Inﬁnitesimally. . For instance. [J i . leads to U (Λ. = pµ |p. 3. P j ] = i δ ij H/c2 . check that U1 U2 corresponds with Λ1 Λ2 ). We have here reinstated c. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . the generators J 2 and J 3 in the case of the rotation group. a) = Λ−1 µ P ν . H] = [J i . . 3.44) (3. K j ] = i ǫijk K k .45) Note that the commutation relations among the generators M µν in Eq. K j ] = −i ǫijk J k /c2 . . Explicitly. ǫ) = 1 − ωαβ M αβ + i ǫα P α . The generators of the translations are the (momentum) operators P µ . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant.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. the generators J± in the case of the rotation group. J j ] = i ǫijk J k . writing the generator P = (H/c. For instance. [P µ .38 could have been obtained in the same way. a)P µ U (Λ. Taking any one of the states in an irreducible representation. H] = i P i . In order to label the states in an irreducible representation we construct a maximal set of commuting operators. P ρ ] = 0. ν (3. a)P µ U † (Λ. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . (3. a) = Λµν P ν or U † (Λ. .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. = −i (g µρ P ν − g νρ P µ ) . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . [K i . P ) in terms of the Hamiltonian and the three-momentum operators.46) [K i . The eigenvalues of these will be the four-momentum pµ of the state. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). P ν ] [M µν .42) (to decide on where the inverse is. They just state that M µν transforms as a tensor with two Lorentz indices. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. (3. [J i . known from non-relativistic quantum mechanics are obtained. because one then sees that by letting c → ∞ the commutation relations of the Galilei group.48) . [K i . other states in that representation are obtained by the action of operators outside the maximal commuting set. the generators M αβ no longer exclusively act in Minkowski space. Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . as part of the set.

54) This will enable us to pick a suitable commuting ’spin’ operator. i ǫµνρσ W ρ P σ . p0 > 0. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. which can thus be chosen spacelike. . (3. 0. .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . .56) . 0. . Massive particles: p2 = M 2 > 0. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). M ǫijk .51) The following properties follow from the fact that W µ is a four-vector by construction. . 2 exp (−isµ W µ ) |p. . Wν ] = −i (gµα Wν − gνα Wµ ) . = = i exp − ǫµνρσ M µν pρ sσ |p.49) (3. Pν ] = [Wµ . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . the third case represents the vacuum. . while the others have no physical signiﬁcance. 2 (3. Wα ] = [Wµ .52) (3. .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. 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.53) (3.55) commute with all generators and therefore are invariants under Poincar´ transformations. (3.

p. .61) one sees that S i (p) i =− W2 . ms .e.65) . (3. 0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. show that W /M can be interpreted as the intrinsic spin. W j (p)] = i M ǫijk W k (p). More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 1. . M2 (3. i (3.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). M2 (3. Together with the four momentum states thus can be labeled as |M. 0). 3. 0. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). The commutation relations [W i (p).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0.64) such that in an arbitrary frame where the four vectors p. 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. . 1 2 (3. p0 > 0. (0. (0. W i (p) = −W · ni (p)). |p3 |). (3. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions).e. 0). n1 (p). 1. 0. 0. (1.57) (i.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. s. 0).59) (3. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 .58) i.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . 2 .. Having made a covariant decomposition. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 0. 1. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.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. M J) with components W i = (M/2)ǫijk M jk .

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 . implying continuous values for the spin (actually for W i = M S i ). But if M = 0 one has a diﬀerent algebra (Eq. Thus we have P 2 . W µ (p) = λ(p) P µ . |p. and gives [W 1 (p). U (Λ.70) as a commuting set with zero eigenvalues for P 2 and W 2 . 3. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). 3. W 2 (3. 1 2 24 (3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). however. i W 2 (p). P µ . A massless particle. a)|0 = |0 . Note that angular momentum does not contribute to helicity as L · p = 0. Exercises Exercise 3. The vacuum: pµ = 0.Groups and their representations The above does actually include the massive case. The algebra of W 1 .13. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). This state is in physical applications nondegenerate and is denoted by |0 . thus. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. Wν ]. W 2 (p)] = = = i M 2 λ(p).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). W 1 (p)] [λ(p). −i W 1 (p). We don’t know of any physical relevance. which can take any integer or half-integer value. W 2 and λ is derived from the general commutation relations for [Wµ .73) (3.67) (3. Thus 2 λ(p) = J ·p . is characterized by its momentum and the helicity.1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq.71) which is called the helicity. |p| (3. The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with W 0 (p) = − 1 ǫ0ijk Mij pk = J · p.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0).67 has a vanishing right hand side). (3. The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). λ . W 2 (p)] [λ(p).72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. (3.68) (3.

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

. Determine the commutation relations for the corresponding operators Ki in quantum mechanics. r2 ) inevitably leads to interaction terms in the boost operators. rj ] = [pi . Exercise 3. H P J K = p 2 c 2 + m2 c4 . indeed assures d K /dt = 0. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. Comment: extending this to more particles is a highly non-trivial procedure. = p. From either of these. if you don’t want to do this in general. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. [ri . show ﬁrst the operator identity [r. ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). they describe r −→ r′ = r − u t while t′ = t. i. This is possible for a single particle. H] + i dt ∂O .5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame. and U (u) p U −1 (u) = p − m u. p×s 1 (rH + Hr) − t p + . [ri . although the presence of an interaction term V (r1 .67 to 3.e. pj ] = [ri . boost invariance leads to a conserved quantity. = r × p + s. there is a one to one mapping) the algebra in Eqs 3. (c) Give the unitary operator for boosts. that’s why many-particle non-relativistic quantum mechanics is ’easy’..e. Do this by showing that the set of operators. in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. (a) In classical mechanics. consistent with the (canonical) commutation relations of a quantum theory. the operator U (u) that gives U (u) r U −1 (u) = r − u t. you might just check relations involving J or K by taking some (relevant) explicit components.6 Consider rotations and translations in a plane (two-dimensional Euclidean space). sj ] = 0. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O.Groups and their representations 26 Exercise 3.69 of the Pauli-Lubanski operators for massless particles. but it can be done. derive the commutation relations and show that the algebra is isomorphic to (i. Exercise 3. the others vanish. sj ] = iǫijk sk . These do not matter in the non-relativistic limit (c → ∞).e. Hint: for the Hamiltonian. momentum and spin operators e satisfy the canonical commutation relations.7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. sj ] = [pi . pj ] = iδ ij and [si . f (p)] = i ∇p f (p). = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. although that in this case is a relation that is found already from the Ehrenfest relation for O = ri . i.

j ′ ). it also follows directly that the Lorentz group is homeomorphic with the group SL(2. Aj ] = [B i . where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞.7) From the considerations above. becoming the deﬁning representations of SL(2. The matrices in SL(2. They will be labeled by two angular momenta corresponding to the A and B groups.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2).9) (4. C) is the group of complex 2 × 2 matrices with determinant one.5) = 1 (J + i K). K = −i . 2 1 (J − i K). 2 (4. Special cases are the following representations: Type I : (j.10) . With this choice of parameter-spaces the plus and minus signs are actually relevant. 0. 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 . respectively. K = i J (A = 0). (j.6) (4. It is simply connected and forms the covering group of L↑ . similarly as the homeomorphism between SO(3) and SU (2). 0) Type II : (0. The group SL(2. U (ϕ) H(η) (4. (4. This tells us how to ﬁnd the representations of the group. C).Chapter 4 The Dirac equation 4.4) (4. 2 2 σ σ J = . B j ] = [Ai . j) K = −i J (B = 0). 2 2 (4. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . i ǫijk B k .1 The Lorentz group and SL(2. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . B j ] = i ǫijk Ak . K = +i . They precisely correspond to the two types of representations that we have seen before.3) (4.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.1) (4. (4.

16) = σ µ Λ R aµ . thus. η → Hη. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. ξ −→ Hξ. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ).1). namely H(η) = exp(η · σ/2). while ±ǫη ∗ transforms as a type-I + spinor.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . C) and.15) (4.13) 2M (E + M ) (exercise 4. C). U σ µ aµ U −1 = σ µ ΛR aµ . This shows that M ∗ ≃ M . In fact.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. σ µ ≡ (1. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.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. ǫ−1 = ǫ† = ǫT = −ǫ). (4.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2.12) ξ → U ξ. 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. H(η) = exp(−η · σ/2). U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. Eqs 4. (4. if there does not exist a unitary matrix S. The following relations for + rotations and boosts are useful. thus. ˜ = σ µ Λ B aµ . 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). This is not true for the two-component spinors in SL(2. however. C) or L↑ . i. Λ(ǫ) ∈ L+ . U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. C). σ). (ǫ∗ = ǫ. η → U η. H ≡ (H † )−1 = H −1 ). −σ). This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . such that M = SM S −1 . are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2.18) As already discussed for SU (2). there exists the matrix ǫ = iσ 2 . one can use a boost to the frame with momentum ˆ ˆ p. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ.19) . (4.11) Considering ξ and η as spin states in the rest-frame. (4. cannot be connected by a Lorentz transformation belonging to L↑ .e. ˜ (4. The Lorentz transformations Λ(M ) and Λ(M ). (4.

+ − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . generators. 4. i. −→ M (Λ) (4. (axial vector).e. i. 4.17. The i angular momentum operators J are represented by σ i /2. (pseudoscalar). This provides the easiest way of seeing ˜ what is happening. We also have seen that the representations (0.g. therefore. but rather the operators W i (p) should be used. as we have seen in section 4. For a particle at rest only angular momentum is important and we can choose ξ(0. m) = η(0. but to L↑ .20) 4.13. where a = (a0 . 0). the aij elements of a tensor will not change sign. C) are suitable for representing spin 1/2 particles.23) u(p. ǫµνρσ will change sign. 0 χm uR H(p) . that have the correct commutation relations. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. ) −→ ( . we have seen that in the rest frame W i (p)/M → J i . (vector). 1 ) and ( 1 .The Dirac equation 29 where The matrix I2 does not belong to L↑ . C). Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. (scalar). A parity transformation changes aµ into aµ . The behavior is the same for classical quantities. m) = χm uL H(p) 0 . −→ −p (vector). etc. Although these operators cannot be used to label the representations.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. Thus M ∗ and M are not equivalent within SL(2. From the deﬁnition of the representations (0. m) = χm . −a). must combine the repe resentations.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. they cannot be connected by a unitary 2 2 transformation. (4.e. It has the same eﬀect as lowering the indices. they can be connected. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . The representing member of the class P− is the parity or space inversion operator Is .1. a) and ˜ a = (a0 . we obtain for the two components of u which we will refer to as chiral right and chiral left components. Taking M (Λ) = H(p). 2 ) and ( 2 . 0) are inequivalent. 2 2 (4. 0) of SL(2. For the spin 1/2 representations of the Poincar´ group including parity we. e. Within the Lorentz group. but by a transformation belonging ↑ ↑ to the class P− . = (4. the boost in Eq.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. the well-known two-component spinor for a spin 1/2 particle. 1 ) and ( 1 . −→ −r (vector).

The ﬁrst indeed is solved. (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 .29) which is an explicit realization of the (momentum space) Dirac equation. (4. 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 = 0.32) suppressing on the RHS the identity matrix in Dirac space.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. . Since ∂µ ∂ν is symmetric.28) which is a covariant (linear) ﬁrst order diﬀerential equation.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).27 is known as the Weyl representation. The explicit realization appearing in Eq. (4. 2 (4.7. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. 4. It is of a form that we also played with in Exercise 2. {γ µ .The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . γ ν } = 2 g µν . ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. p (4. = H −1 (p). We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. (iγ µ ∂µ − M ) ψ(x) = 0. γ } ∂µ ∂ν + M 2 ψ(x) = 0. = (4. 0 H 2 (p) uR uR . (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. which in general is a linear equation in pµ . one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately).24) (4. this can be rewritten 1 µ ν {γ . (4. (4.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. We will discuss another explicit realization of this algebra in the next section. From this one obtains the Cliﬀord algebra for the Dirac matrices. Applying (iγ µ ∂µ + M ) from the left gives (4. The general requirements for the γ matrices are thus easily obtained.33) deﬁne for a four vector a the contraction a = aµ γµ / .

g.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. (4.36) (4. γ ν } = 2 g µν . This problem was solved by Dirac through the introduction of a negative energy sea. opposite charge). The second problem encountered before. This can most easily be seen in the particle rest frame. E = +M (twice) and E = −M (twice). Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. From the vacuum a particle can be removed.34) for the adjoint spinor ψ ≡ ψ † γ0 .1: The Dirac sea of negative energy states and antiparticles. (4.35) is indeed positive and j 0 can serve as a probability density. (4. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). the one of the negative energy states. Relying on the Pauli exclusion principle for spin 1/2 particles. The density j 0 = ψγ 0 ψ = ψ † ψ (4. The Dirac sea forms the vacuum. 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. This hole forms an antiparticle with the same mass. . 4. remains.3 General representations of γ matrices and Dirac spinors {γ µ . .27). 4. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4.38) 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 . one sees that there are two positive and two negative eigenvalues. but with properties such that it can be annihilated by the particle (e.37) Using the explicit form of γ in the Weyl representation (see Eq. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.

S −1 .R. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . which in the case of the Weyl representation are just the upper and lower components.3) .41) (4. = S(γµ )W. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . (γµ )S.46) (4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .R. γ µ } = 0 and explicitly one has 0 1 1 . −→ Sψ. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. (4. γ 2 ].43) S=√ . −→ Sγµ S −1 . γ σ ]. 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 . For example. while L−1 γ µ L = Λµ γ ν . For instance in the Weyl representation (but valid generally). is 1 1 1 (4.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 .40) (4.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .47) (4. that project out chiral right/left states.R.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. 4! (4.9 and 4. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 . γ 0 ]. =ρ ⊗1= 0 −1 3 (4. 2 (4. We note that ψ → Lψ and ψ → ψ L−1 .R.45) (γ5 )W. (γ5 )S. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. 0 (4.48) are projection operators. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . the rotation and boost generators in Weyl representation in Eqs 4. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k .

∂ (4. One has ψ c = −ψ T C −1 . in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . x = (t.53) (4.) and all representations connected via a real (up to an overall phase) matrix S. T (4. ψ= (4.g.57) (4.52) η ξ one can apply space inversion. −iσ 2 0 −ǫ = . ∂ with the solution ψ c (x) = ηc Cψ (x).55) (4. is a left-handed spinor. for instance.R.R. reverse the charge of the particle.58) where ηc is an arbitrary (unobservable) phase.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. The latter implies that the conjugate of a right-handed spinor. ψ= (4. In S. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . 0 −ǫ 0 0 −ǫ 0 . which is again a solution of the Dirac equation. Explicitly.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.g.54) e. γ5 ] = 0. Some properties of C are C −1 = C † and C T = −C. usually to be taken unity. (or W. 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. T (4.56) (i/ − M ) ψ c (x) = 0. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. This symmetry does not change the spin 1/2 nature.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. but it does. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. = 0 ǫ iσ 2 (4.6). called charge conjugation. = = 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.R. As with parity we look for a transformation. In any representation a matrix C exist.59) −→ ψ c = C ψ = η ǫ ξ∗ . ∂ (4. C is real and one has C −1 = C † = C T = −C c and [C.R. x) → x = (t. that brings ψ → ψ c . ψR . (i/ − M ) ψ(x) = 0. e. such that T Cγµ C −1 = −γµ . (C)W. (C)S. parity.

s)u(p. p (4. In an arbitrary frame one has s · p = 0 and s(p) can e. s ) = 0. the best representation to describe particles at rest is the standard representation. s) e−i k·x b(k. be obtained by a Lorentz transformation.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions.The Dirac equation 34 4. s) and C v T (p. (4. s ) = v † (p.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. ψ(x) = v ±s (p) ei p·x . The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. (2π)3 2Ek (4.62) where χs are two independent (s = ±) two-component spinors. s)¯(p.64) (4. s) = v(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.68) (4. s′ ) = 2Ep δss′ . s). ψ(x) = u±s (p) e−i p·x .70) = Ps = ˆ 0 1∓σ·s 2 2 . in which γ 0 is diagonal (see discussion of negative energy states in section 4. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. with E = Ep = + p2 + M 2 . ¯ ¯ † ′ u (p.1). s) = Ep + M σ ·p Ep +M χs . s)v(p. s) + v(k. 2M 2M −/ + M p v(p. s) = u(p.61) σ·p −(Ep + M ) There are also two negative energy solutions. s ) = v (p. s)v(p. s) ei k·x d∗ (k. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. ⇔ (/ + M ) v(p) = 0. s) . p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. the spinors satisfy C uT (p. ⇔ (/ − M ) u(p) = 0. the spin polarization vector in the rest frame is the starting ˆ point. s′ ) = 2M δss′ .65) (4. s)u(p. v(p.63) u(p.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s′ ) = −2M δss′ . Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s). s)u(p. In that frame is a spacelike unit vector sµ = (0. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s)¯(p. 2M 2M (4.69) = − s In order to project out spin states. ¯ (4. ¯ ′ ′ Explicit solutions in the standard representation are χs .g.60) ψ(x) = 0. v (p. ∂ −iσ · ∇ −i ∂t − M (4. where u satisﬁes −σ · p Ep − M u(p) = 0. The ¯ ¯ solutions are normalized to u(p. s) v = . s)v(p. s) = u(p. s) u = ..

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

s)γ µ u(k ′ . (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!.75) Lµν (k. s′ )) (¯(k. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd.76) ul (k ′ . (6) Tr(γ5 ) = 0. s′ )¯i (k ′ . s)(γ ν )kl u u s.s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. In principle one can take a representation and just calculate the quantity u(k. a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ .The Dirac equation 36 4. s′ ) ¯ u s. s)γ ν u(k ′ .5 γ gymnastics and applications An overview of properties can be found in many text books.g µρ g νσ + g µσ g νρ ). (8) σµν ≡ (i/2)[γµ . (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. s′ )γ µ u(k. s)γ µ u(k ′ . s′ )) .g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . s).s which appears in quantum electrodynamics calculations (see Chapter 10) for the emission of a photon from an electron changing its momentum from k to k ′ (k 2 = k ′2 = m2 ). (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. u u (4. ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ . k k 2 . Some elementary properties are: Properties of products of γ-matrices: (1) γ µ γµ = 4 (2) γ µ γ ν γµ = −2 γ ν or γ µ /γµ = −2 a. however. s′ )) = u(k ′ . It is.) (2) Tr(1) = 4. (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 . (Use (γ5 )2 = 1 and pull one γ5 through. s′ )(γ µ )ij uj (k. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. s)γ µ u(k ′ . namely the calculation of the quantity 1 ∗ (¯(k. we will give one example.) In order to show the use of the above relations. more convenient ¯ to use the projection operators introduced earlier. k ′ ) = 2 ′ s. γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. s)¯k (k. u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ . For this we ﬁrst have to prove (do this) that (¯(k.s′ ∗ (4. (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ). s′ )(γ µ )ij uj (k. (5) γ µ γ λ γ ν γ ρ γ σ γµ = 2 γ σ γ λ γ ν γ ρ + γ ρ γ ν γ λ γ σ . s)(γ ν )kl ul (k ′ . s′ ). etc. s)¯k (k.

γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. −x) is also a solution of the Dirac equation. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . (4. s). H(p) = exp η·σ 2 = M +E+σ·p . They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . s)¯(p.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . s) and v(p.g. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M.4 Show that in P+ ≡ s=±1/2 u(p.77) Exercises Excercise 4.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. Excercise 4. Show that the equality holds. 1 i −i S ρσ = − [γ ρ . x ˜ .5 Show starting from the relation {γµ . 4. ˜ Exercise 4. of which only the traces of four and two gamma-matrices are nonzero. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . γ σ ] = σ ρσ . Exercise 4. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.6 Apply space-inversion.1 Prove Eq. a 4 a · b. s) u p /+M = 2M 2M both sides are projection operators. where x = (t.g. by letting them act on the four basis spinors u(p. 4 2 Exercise 4. 0 if n is odd. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). x → x. Do this e. e.13.

For such a cavity nµ = (0. M R = 1. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.66 MeV. we want to conﬁne this solution to a spherical cavity with radius R.511 MeV). namely i / ψ = eiαγ5 ψ.8 × 10−6 . Show that this n condition is suﬃcient to guarantee that nµ j µ = 0.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). ν p Note: how such parametrizations work will be explained in chapter 8. Calculate for now just |M |2 .e. 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 ). Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. you need to determine the corresponding conﬁnement condition for ψ. is a stationary solution of the (free) Dirac equation. 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. i. 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.57 MeV.5 and M → ∞. n It turns out that there is actually a whole family of conditions that provide conﬁnement. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. mµ = 105. r). Note: To prove this. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . . me = 0. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. We will not pursue this chiral freedom at this point and take α = 0. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. Do you understand why? The calculation of the life-time is completed in chapter 11. Calculate the value of k for M R = 0. (b) Next. The pion has spin 0. which is the most important factor needed to ﬁnd the decay width (inverse lifetime).7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. the decay would even be forbidden. Excercise 4. there is no current ﬂowing through the surface of the cavity. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal).The Dirac equation 38 Exercise 4.

The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. These are solutions of the Klein-Gordon equation with in addition a condition. In an arbitrary frame ǫµ (k) with λ = 0. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. with ∂µ V µ = 0 (Lorentz condition). there are then indeed three independent possibilities.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation.5) −E B2 . where k = k0 = (M. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . just as the two-component Dirac spinors did for spin 1/2. (2 + M 2 )Vµ = 0.4) λ=0.7) .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. µ (2π)3 2E µ (5. These can describe spin 1 states in the rest-frame. 39 (5.Chapter 5 Vector ﬁelds and Maxwell equations 5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0.2) (5. ±(1) are obtained by boosting the restframe vectors.1) The Lorentz condition implies that k µ ǫµ (k) = 0. ǫλ ). λ) . In the restframe.6) where Aµ is the four-vector potential (5. namely ǫµ (k0 ) = (0. λ) + ǫ(λ)∗ (k) eik·x c∗ (k. M2 (5. (5. The vectorﬁeld is therefore (λ) suited to describe spin 1.1 Fields for spin 1 In section 3.± 5. 0). A).

AP (x) = Aµ (˜). (5.12) (5. −i. in accordance with what we found in section 3. if in Eq. Although the Lorenz condition is a constraint. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. |k|).16) (5. 0.8) (5.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. As remarked above. ˜ x Note that via parity transformation one obtains another solution. in which case one has 2Aµ = jµ . it will vanish at all times. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). Furthermore. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. 40 (5.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. It states the absence of magnetic charges and Faraday’s law. 5. (5. One possible choice is the gauge A0 = 0 (5. . shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. i. (5.14) of which the solutions give the Li´nard-Wiechert potentials. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . Under a gauge transformation Aµ −→ Aµ + ∂µ χ.11) This equation is not aﬀected by a gauge transformation. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0.5. The equation in vacuum.17) (radiation. 0. For electrodynamics one has furthermore the freedom of gauge transformations.10) the electric and magnetic ﬁelds are unchanged. ∂µ Aµ = 0. ∓1. (5. only two independent vectors remain ǫ(±) µ (k) = (0. (5. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ .e.15 the d’Alembertian is replaced by 2 + M 2 . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. transverse or Coulomb gauge).13) Hence. Therefore if k µ = (|k|. 2Aµ = 0. if ∂µ Aµ = 0 for large times |t|. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld.15) (5. Taking the divergence of this equiation. moree over. 0).9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. for M = 0. j) the Maxwell equations read ∂µ F µν = j ν .

0 . 5. It is however.19) are gauge invariant. 0). ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. but A = 0 (hence there is structure in the vacuum). To be precise. λ L − λ λ or ∆x = (L− λ/d)δ. ˆ B = 0. . In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. cos ϕ. = exp −i The phase diﬀerence between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. . Inside solenoid: Outside solenoid: B = B z. A= By Bx Br ϕ= − ˆ . In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). but where E = B = 0. 0) and ϕ = (− sin ϕ.0 . This phase diﬀerence is actually observed. where Aµ = 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 . The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . may look a bit akward.Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. sin ϕ. The situation. 2 2 2 BR2 BR2 y BR2 x A= . Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. This causes an interference pattern as a function of ∆x. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions).1: The Aharonov-Bohm experiment 5. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). (5. while the electromagnetic ﬁeld Aµ is not. illustrated in ﬁg. nevertheless. This occurs in the region outside the solenoid.3 The electromagnetic ﬁeld and topology∗ The electric and magnetic ﬁelds E and B as appearing for instance in the Lorentz force on a moving charge.1.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. F = e E + ev × B (5. .

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

5. ∂t ∂B = 0. .9 explicitly give the equations ∇ · E = ρ. ∂t ∇ · B = 0.Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5.8 and 5. ∇×B =j+ ∇×E+ ∂E .1 Show that Eqs.

t1 (6. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.3) (6. These equations can be obtained from a lagrangian using the action principle.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. ∂x (6. x) ˙ (6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. ∂x ˙ (6.8) 44 . x) = K − V = ˙ 1 mx2 − V (x). x (t) = x(t) + δx(t).6) The ﬁrst term leads to the Euler-Lagrange equations. d dt ∂L ∂x ˙ = ∂L . thus t′ = t + δτ. ˙ p= ∂L .1) and as an example the lagrangian L(x. As an example.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 . recall classical mechanics with the action t2 S= t1 dt L(x. ˙ 2 (6.1 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).Chapter 6 Classical lagrangian ﬁeld theory 6.5) with ∆x(t) = δx(t) + x(t) δτ .2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. (6.

10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ).18) 1 Taking a functional derivative.t2 ) d3 x . R (6. ˙ x The second term in Eq. . . i. ∂µ φ(x)) = R d4 x L (φ(x). = φ(x) + δφ(x) (6.11) λ=0 of which p and H are the simplest examples related to space and time translations. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . ∂µ φ′ .e. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). indicated with δF [φ]/δφ should pose no problems. .t1 ) (6. x). + ∂L ∆x − H δτ ∂x ˙ t2 . (R3 . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). . with ∆φ(x) = δφ(x) + (∂µ φ)δx . (6. dσµ . ∂µ φ. but which in 3 dimensions includes conserved quantities related to rotations and boosts. . In general any continuous transformation. . . ∂µ φ(x). Variations in the action can come from the coordinates or the ﬁelds.Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. = ∂R (R3 .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ .9) which is done because the ﬁrst term (multiplying ∆x. think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 .2. Consider a lagrangian density L which depends on these functions. t1 ) and R3 . λ=0 (6. L (φ(x). . δ(∂µ φ) (6. 6.15) d4 x′ L (φ′ .17) Furthermore the variations δφ and δ∂µ φ contribute to δS. H is a conserved quantity. their derivatives and possibly on the position. . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). ˙ (6.6 can be rewritten as δS = . which in classical mechanics vanishes at the boundary) does not play a role.13) (6. t1 (6. t2 ). . We will come back to it in a bit more formal way in section 9. ∂µ φ(x)). x′ ) − d4 x L (φ. − d3 x . ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. x) ≡ p x − L .14) (6. 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 . The resulting variation of the action is δS = R µ ′ = xµ + δxµ .12) Here R indicates a space-time volume bounded by (R3 . The hamiltonian H is deﬁned by H(p.

(6. (6.23) (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. ∂ (6.27) (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. 2 ∗ (6.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2.19) δ(∂µ φ) δφ 6. One easily obtains (2 + M 2 )φ(x) = 0.21) which diﬀer only by surface terms. Using the variations in ψ (in the symmetric form). The integrand of the ﬁrst term must vanish. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. leading to (2 + M 2 )φ(x) = 0. the second term is irrelevant. 2 (6.29) .20) (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.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.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. leading to the Euler-Lagrange equations. (2 + M )φ (x) = 0.25) (6. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ.2 Lagrangians for spin 0. 2 ψ = 0.24) (6.

34) (6. or equivalently separate the ﬁeld into right. We note that of which the left part coincides . 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.g. ∂ ∂ 2 2 (6. Peshkin and Schroeder or Exercise 12. and thus (β ∗ α)∗ = α∗ β = −βα∗ . Using the twospinors ξ and η.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).and lefthanded ones.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. 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. 2 αβ = −βα. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 .g.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL .36) (6. 1 M η † ǫη ∗ − M ∗ η T ǫη . In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.38) (6. The reasons for Grassmann variables will become clear in the next chapter.7). (6.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . ψL ≡ (ψL )c = C ψL = 0 (6. but note the factor 1/2 as compared to the Dirac lagrangian. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . The Majorana case is in fact more general.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. (6. 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 . . the mass term in the Dirac lagrangian 6. (6. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . which comes because we in essence use ’real’ spinors. that the lagrangian separates into two independent parts for M = 0.32) There exists another possibility to write down a mass term with only one kind of ﬁelds. namely2 LM (Majorana) = + (6.31 is given by LM (Dirac) = −M ξ † η + η † ξ .31) showing e. 2 (6.

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

56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . From Noether’s theorem one sees that the current Θµν ǫν is conserved. 1 (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .49) These currents are conserved as discussed already in chapter 2. (6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν .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. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). In the next section we will see the quantity show up as the charge operator. H = d3 x Θ00 (x). (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). . 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. ↔ (6. e.g.3) j µ = i e φ∗ ∂µ φ. (6. including e the space-time translations and the Lorentz transformations. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. 6. 2 (6. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν .Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.54) (6. (6.53) These are the energy and momentum. (6. = 0. we obtain (since δxµ = 0). t) is the conserved charge (Q = 0). Summarizing. The integral over the zero-component.51) (6.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space).4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters. ˙ Q = d3 x j 0 (x.50) (6.

∂µ T µν (6.68) (6.67) i. (6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . Any lagrangian containing derivatives.59) (6. (6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. 6. (6.63) µν = ∂µ Θ . In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric.3 we have seen global transformations or gauge transformations of the ﬁrst kind. (6.61) A ﬁnal note concernes the symmetry properties of Θµν . Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). e. in which the U (1) transformation involves an angle e Λ. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). Deﬁning T µν = Θµν − ∂ρ Gρµν . is not invariant under local gauge transformations. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. ∂µ φ(x) + i e ∂µ Λ(x) e (6.65) M µρσ = T µρ xσ − T µσ xρ .e. 2 (6.66) In section 6.70) . where H ρµν is the quantity appearing in M ρµν . independent of x. however.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).69) i eΛ(x) φ (x). • Vector ﬁeld Aµ : −i Sρσ = aρσ . (6. In general this tensor is not symmetric.6) T µν = T νµ . The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. φ(x). 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 .64) (6. one has a tensor that satisﬁes (exercise 6.58) (6.3).g. the angle of the transformation depends on the space-time point x. −i eΛ(x) ∗ i eΛ(x) (6.

74) the behavior which we have encountered before as a gauge freedom for massless vector ﬁelds with the (free) lagrangian density L = −(1/4)Fµν F µν . The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ.73) (6.77) 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). Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. As an example consider the Dirac lagrangian.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.72) (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 . 4 (6. Dµ φ) − 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). 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. −e j µ Aµ = −e ρφ + e j · A. ∂ 2 4 (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. (6. D ∂ A (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ.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 . (6. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. 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. ∂µ φ) =⇒ L (φ.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.78) . (6. 1 L (φ.

[iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). 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.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . i. ↔ Exercise 6.79) Exercises Exercise 6.1 (a) Show that the Klein-Gordon equation for the real scalar ﬁeld can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 . φ → eiΛ φ. Exercise 6. and deduce what is the charge of the antiparticle as compared to a particle. giving the Maxwell equation coupling to the electromagnetic current. ∂µ F µν = j ν . ∂µ F µν = j ν . 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. Exercise 6. .2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. where j µ = e ψγ µ ψ. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. give the equation which is satisﬁed by ψ c = Cψ . −eψγ ν ψ.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.e. (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 . It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic ﬁeld takes the value gs = 2. .Classical lagrangian ﬁeld theory 53 Exercise 6. Hint ≡ −gs µe · B = −gs Qe s · B. A) (see also exercise 6. Comment: this relation is known as the Pauli equation. 2me Exercise 6. 2M 2M with Enr ≡ E − M ≪ M . in order to show that Θρσ − Θσρ = ∂µ H µρσ . ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. use ∂µ T µν = 0 and ∂µ M µρσ = 0. This relation is known as the Pauli equation.4.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 .6 (optional) Prove the properties of the tensor T µν in section 6. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. t) = Enr ψu (r. t).

An immediate generalization for ﬁelds can be obtained by considering them as coordinates. Quantization of the theory is achieved by imposing canonical commutation relations between q and p.Chapter 7 Quantization of ﬁelds 7. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. Further. dλ Examples are dO = [O.11) found through Noether’s theorem.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). ˙ L(q. t). (7. (7. labeled by the position. in our example p = mq. B]P = dB dA dA dB − . [q. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. Q]P . p] = i. is the Poisson bracket [A. there are conserved ˙ ˙ quantities Q (Eq. q) = p q − L(q. q(q. dq dp dq dp (7. (7. 1 qx (t) = 3 d3 x φ(x. q) also considered in the previous chapter. 6.2) H(p. however.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t).5) ∆ x ∆3 x 54 . p]P = 1.3) For variations δO(p. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. q) = ˙ 1 mq 2 − V (q). ˙ 2 (7. starting from the lagrangian L(q. p]P . namely [q. The bilinear product for the conserved quantities. H]P dt and dO = [O. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q.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.

a) = Rj (Λ−1 ) φj (Λ−1 x − a). ∇φ) (7. for the scalar ﬁeld theory. t)] = 0. a) φi (x)U (Λ. ˙ δ φ(x) (7.15) with furthermore the relations [φ(x. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). the discretization procedure above is an explicit example.7) d3 x L (φ(x). The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). t). t). • Poincar´ invariance e The full variation of the ﬁelds. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . qx+∆x ) . t). ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x.11) (7. Φ(f ) = d4 x φ(x) f (x). ∆φ. As an example. 2 2 2 (7. (7. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). qx . t)] = i δ 3 (x − y).13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. Sometimes it is useful to keep in mind that. 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.8) (7.Quantization of ﬁelds 55 etc. φ. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 .12) (7. (7. φ(y. formally.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . t)] = [Π(x. albeit with ’sharp’ functions. ∂µ φ(x)) = ˙ d3 x L (φ. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. Π(y. ∂ qx ˙ ∂ qx ˙ (7.14) In fact. discussed in the section 6.16) . for symmetry transformations.2. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. (7.6) ∆3 x L x (qx . Π(y. i U † (Λ. The essential conditions to consistently quantize a theory then are the following.

Q] = [P. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). φ(y)] = 0 if (x − y)2 < 0. 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. a] = −a.24) It is straightforward to ﬁnd the commutation relations between N and a and a† . i(xµ ∂ν − xν ∂µ )φ (x) + i (7. P ] 2 (7. a† ] = 0. a† ] = a† .22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. j 2 one has [φi (x). Microscopic causality implies local commutativity. (7. The measurements cannot inﬂuence each other or [Φ(f ).17) (Sµν )i φj (x). a] = [a† . a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . (7. P ] = i. Mµν ] = i i∂µ φi (x). Φ(g)] = 0. sometimes referred to as second quantization.18) with S µν given in 6.4.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. P ] = 0 (7.19) 7. [φ(x).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.23) The hamiltonian in this case can be expressed in the number operator N = a† a. 2 2 (7. j (7. Pµ ] = [φ (x). [a. [Q. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . [N.Quantization of ﬁelds 56 or if U (Λ. and [N.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. 2 i ωµν (S µν )i . In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . (7. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7.25) . ω a† a + 2 2 ω i − [Q. a† ] = 1.

31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). † √ a and a act as raising and lowering operators. is equivalent with [a(k). φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . 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 . The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). Π(x )]x0 =x′0 = 0. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.g.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . but one needs to realize that in that case k 0 = Ek = k2 + M 2 .30) [φ(x). φ(x )]x0 =x′0 = [Π(x).26) a(k) e−i k·x + a† (k) ei k·x . and (7. a(k ′ )] = [a† (k).3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.like’ commutation relations between a(k) and a† (k ′ ). a† (k ′ )] = 0.32) (7. (7.28) (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . 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.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). and we see that a state |0 must exist for which N |0 = a|0 = 0. Note that we will often write a(k) or a† (k). e.e.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. (n − 1) a|n † i. [φ(x). d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .33) .e. 7. which represents the ’number of particles’ with momentum k. the real scalar ﬁeld φ(x) becomes d3 k (7. (2π)3 2Ek (7. i.

(2π)3 2Ek (7.34) where V = (2π)3 δ 3 (0) is the space-volume. . deﬁning particle states |k = |k (with positive energy. . (2π)3 2Ek (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). i. d3 k |k k|. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . is solved by subtracting it as an (inﬁnite) constant. d3 k a(k) e−i k·x . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . (2π)3 2Ek d3 k a† (k) ei k·x . (2π)3 (7. This term will be adressed below.35) where the vacuum contribution disappears because of rotational symmetry.40) This procedure is known as normal ordering. 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).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.41) (7.44) p|φ− (x)|0 = ei p·x .45) (7. |k = a† (k)|0 .e. .36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). k 0 = Ek = k2 + M 2 ). (2π)3 2Ek (7.46) (7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. writing all annihilation operators to the right of the creation operators.42) (7. n1 ! n2 ! (7.37) (7. . a(k)|0 = 0. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. |0 .39) The problem with the vacuum or zero-point energy. .43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y). which now contains an inﬁnite number of oscillators. (2π)3 2Ek (7.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. and multiparticle states |(k1 )n1 (k2 )n2 . The rest of the states are then obtained from the groundstate via the creation operator.

i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (2π)3 2Ek ≡ i∆− (x − y). (iii) ∆(0. d3 k = (2π)3 2Ek = (7. x) = 0 and hence ∆(x) = 0 for x2 < 0.e. φ(y)] consider µν [φ+ (x). 2π (7.18).55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .51) (7. = − (2π)3 2Ek (7. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (2π)4 (7. The last item to be checked for the scalar ﬁeld are the causality condition.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). d3 k ei k·(x−y) = −i∆+ (y − x). i. 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 ).17) and (7.48) (7. (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . ∆(x) = − ǫ(x0 ) δ(x2 ).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.50) or as integrals over d4 k.49) (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. φ+ (y)] d3 k e−i k·(x−y) . In order to calculate [φ(x). (7. (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. that they satisfy the required commutation relations in Eqs (7.52) The result for the invariant commutator function is [φ(x).53) t=0 . φ− (y)] ≡ i∆+ (x − y).56) = −δ 3 (x).

Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). we will consider it as a repetition of the quantization procedure. extending it with the charge operator and the introduction of particle and antiparticle operators. (2π)3 2Ek (7.57) (7. The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.65) 7.63) i. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . ↔ (7. 2 From the lagrangian density (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). (2π)3 2Ek (7.61) (7. The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) . = ∂{µ φ∗ ∂ν} φ − L gµν .e.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ. a† (k ′ )] = [b(k). φ† (y)] = i∆(x − y). which is equivalent to the relations (only nonzero ones) [a(k). 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) .66) .59) (7.Quantization of ﬁelds 60 7.62) (7. (2π)3 2Ek (7. (7.60) = i φ∗ ∂µ φ.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).

72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . s)v(k. s). The quantized ﬁelds are written ψ(x) = s d3 k b(k. interchanging creation and annihilation operators.73) (7.69) (7. † {ψi (x).70) where the last line is obtained by using the Dirac equation.71) (7. (2π)3 2Ek d3 k b† (k. ↔ i ψγµ ∂ν ψ − 2 (7. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s)vj (k. s) ei k·(x−y) x0 =y 0 .74) (7. s)ei k·x . {b(k.e. s′ )} = {d(k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. (2π)3 2Ek (7. (7.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). (7.77) Note that achieving normal ordering. The solution is the introduction of anticommutation relations. s)¯(k.78) (7. s)u† (k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . (2π)3 2Ek (7.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. b† (k ′ . . s) e−i k·(x−y) j s † + vi (k.75) (7. 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) . ∂ 2 ↔ (7. s)ei k·x + d(k. s)e−i k·x + d† (k. u v (2π)3 2Ek (7. s)¯(k.67) i ψ / ψ − M ψψ gµν . s)u(k. s)e−i k·x . s). i. jµ Θµν = = ψγµ ψ. d† (k ′ .Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained.

80) 3 = δ 3 (x − y) δij .83) From the lagrangian density the canonical momenta are = = (7. as the vanishing of Π0 induces a constraint. we would have obtained [ψi (x). ˙ δ Ai (7. but has the problem of being noncovariant. ψ j (y)} at arbitrary times one has {ψi (x). 4 Π0 Πi δL = 0. 4 2 This gives the equations of motion discussed before.87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. For the scalar combination {ψi (x). 7. −λ(∂ρ Aρ ).6 The electromagnetic ﬁeld 1 L = − Fµν F µν .86) (7. ∂ (7. ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem).88) . 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. ∂ (7. ˙ δ A0 δL = F i0 = E i . ψ j (y)] = (i/x + M )ij i∆1 (x − y). but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y).81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.84) (7.85) which reﬂects the gauge freedom.82) where i∆1 = i∆+ −i∆− . one has † {ψi (x). (7. (7.

λ)e−i k·x + c† (k. pj ]P . λ) − c† (k. The canonical equal time commutation relations are [Aµ (x). j = 1. where Q is the conserved quantity in Eq. ˙ ˙ .93) which does exhibit problems with the time-like photon. 6. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). 7. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. In fact the commutation relations imply ˙ [Aµ (x). and [ℓi .92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. (d) Use Poisson brackets to calculate q and p.11.Quantization of ﬁelds 63 for physical states |A and |B .89) with four independent vectors ǫµ . µ λ=0 (7. which reproduce the Hamilton equations. Aν (y)] = [Πµ (x). (7.95) Exercises Exercise 7. λ). for which it is suﬃcient that ∂µ Aµ |A = 0. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ).90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x).e. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. c† (k ′ . It gives 3 k µ ǫ(λ) (k)c(k. what are [ri . and are equivalent with [c(k. µ λ=0 (7.3) satisfy the conditions of a Lie algebra. 0. rj ]P . containing a time-like photon. 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. q) that dO/dλ = [O. These problems are solved by the Lorentz constraint between physical states given above. λ)|A = 0. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. (7. 2. λ)ei k·x . i. i. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0.91) (7. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. (b) Proof for a quantity O(p. [ℓi . (7. λ)c(k. Πν (y)]x0 =y0 = i gµν δ 3 (x − y).1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. 0) . pj ]P .e. a longitudinal photon and two transverse photons.94) Choosing k µ = (|k 3 |. ′ (7. Πν (y)] = 0. 0. 0)c(k. Q]P .

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

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

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

x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. the above operation reverses the sign of λ). −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.4) (8. −λ).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. λ) e−ik·x − d† (k. . λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. one can also consider the scalar ﬁeld and vector ﬁelds.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).13) ψ(x) −→ ψ c (x) = ηC C ψ (x).11) (8. m) γ0 v(k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. γ x (8.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). if helicity λ is used instead of the z-component of the spin m. m) = = ˜ u(k.9) (8. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. (i˜ µ ∂µ − M ) ψ(˜) = 0. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). 8. iγ µ† ∂µ − M ψ(˜) = 0.6) (check this for the standard representation. (2π)3 2Ek (8. ˜ Pop d(k.7) = λ d k x x ηP b(k. −λ) e−ik·x − d† (k.10) i. m). x The latter behavior of the vector ﬁeld will be discussed further below. λ)u(k. λ)γ0 v(k. λ)γ0 u(k. x (8. In the same way as the Fermion ﬁeld. λ)v(k. λ) eik·x . −λ). −λ) e−ik·x − d† (k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k.3) (8. ˜ −v(k. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. λ)u(k. λ) P −1 = −η ∗ d(k. −λ)v(k. λ) e−ik·˜ + d† (k. m). λ)v(k.e.5) (8. −λ)u(k. λ) Pop = ηP b(k. op P (8.

as discussed in section 4). antilinear operator is anti-unitary if A† = A−1 . λ) e−ik·x + b† (k. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0.19) Cop d(k. λ). x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. λ) Cop = ηC d(k. λ)C uT (k. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . λ) eik·x . r).17) 3 = λ This shows that −1 Cop b(k. i. m) ¯ = = v(k. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. λ). x (8. Top |φ = φt | = (|φt )∗ . λ) e−ik·x + b† (k. 8. x (8. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. u(k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.21) where A is a 4 × 4 matrix acting in the spinor space. Norm conservation requires Top to be anti-unitary2 . x x (8. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). λ)u(k. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . x iγ µT ∂µ − M ψ(−˜) = 0.18) (8.15) (where one must be aware of the choice of spinors made in the expansion.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). . m). λ)v(k.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .Discrete symmetries 68 the latter being also a solution of the Dirac equation. r) −→ −˜µ ≡ −xµ = (−t. it is antilinear1 . (8. λ) −1 Cop = ∗ ηC b(k.3 Time reversal xµ = (t. we start with the complex conjugated x Dirac equation for ψ. The (time-reversed) Dirac equation becomes.e. The same relations hold for helicity states. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. As time reversal will transform ’bra’ into ’ket’. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8.16) d k ηC d(k. (8. λ)C v T (k. m). (2π)3 2Ek (8. m) ¯ C v T (k.14) (8.

32) (8. λ) T −1 = η ∗ d(k. λ)v ∗ (k. λ |¯.25) (check this for the standard representation). λ)u∗ (k. op T (8. λ). −p.28) (8. λ). Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. λ) eik·x + d† (k. −p. λ)v ∗ (k. λ). λ| ¯ C |¯. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.Discrete symmetries 69 Table 8. (8. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). p.34) .24) (8. λ)iγ5 Cv(k. there are 16 independent of these bilinear combinations. λ)u∗ (k. p. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ a P |a.31) (8. −λ a T a. λ) i γ5 C v(k. λ| a. Since there are 16 independent 4 × 4 matrices.29) In table 2 the behavior of particle states under the various transformations has been summarized. p. ∗ ˜ v (k. λ) Top = ηT b(k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. a state |a. ˜ = 0|b(k)|A 8. λ)u∗ (k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ). λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ) eik·x + d† (k. ˜ Top d(k. λ) e−ik·x (2π)3 2Ek (8. x (8. λ) eik·x + 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. The result is = = λ (8. λ)v ∗ (k.26) d k x x ηT b(k.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ)iγ5 Cu(k. λ) eik·˜ + d† (k. −p.33) (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯).30) (8. −λ |¯. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . p. λ a |a.

41) 2M 2M . (q 2 ). p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). e. As an example consider the vector current for a point fermion. and T. (8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. p′ . p′ |V µ (x)|p = u(p′ )Γµ (p′ .38) (8. C. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . F. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .37) (8.Discrete symmetries 70 Table 8.35) The x-dependence can be accounted for straightforwardly using translation invariance. of which the expectation value between momentum eigenstates can be simply found. C. 2M (8. s′ |V µ (x)|p. called form factors. ¯ ′ ′ (8. 8. s . σ µν . as well as under the combined operation Θ = PCT (see Table 3).g. for the vector current V µ (x). For instance for nucleons one has Γµ (p′ . ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. γ µ . 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 . In many applications the expectation values of currents are needed.40) where the coeﬃcients are functions of invariants constructed out of the momenta. γ ν ]. 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).. 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 ). The 16 combinations of Dirac matrices appearing above are linearly independent.5 Form factors Currents play an important role in ﬁeld theory. (8. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).3: The behavior of the independent bilinear combinations of fermi ﬁelds under P.39) where Γµ (p′ . Note that the lagrangian density L (x) → L (−x) under Θ. p′ |V µ (0)|p = u(p′ )γ µ u(p). (q 2 ). F. in this case only q 2 = (p′ − p)2 . p)u(p) e−i (p−p )·x . p) can be built from any combination of Dirac matrices (1. γ5 γ µ or γ5 ). ¯ In general the expectation value between momentum states can be more complicated.36) (8..

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

(b) Show that in the interaction with the electromagnetic ﬁeld.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 0.Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . Exercise 8. (c) Show that if a term of this form would exist. (c) Show that hermiticity of the current requires that F1 en F2 are real. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . Exercise 8. 0. 0. 0. 0. GE GM = F1 − Q2 F2 .2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. p = (E. 4M 2 = F1 + F2 . (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). |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 . p′ = (E. −|q|/2). σµν q ν F3 (q 2 ) 2M . |q|). 0.

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. t ′ |q . ≡ 0. in which the states are time-independent. t ′ QH (t′ )|q ′ . (9. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t ≡ q|q. (9. and the operators are time-dependent. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. QH (t)|q.5) (9.7) 73 . (9. t0 ) ∂t (9. |ψH (t) = |ψH .1 Time evolution as path integral U (t. |ψS (t) = U (t. [AH . t0 ). in which the operators are time-independent. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian.2) (ii) Heisenberg picture. t0 ) AH (t0 ) U (t. t0 )|ψS (t0 ) . t′ . t′ ≡ q ′ |q ′ .4) ∂ U (t. Consider the two (time-independent) Heisenberg states: |q. H]. t .3) (9. t0 ) = H U (t. AS (t) = AS and the states o are time dependent. (9.Chapter 9 Path integrals and quantum mechanics 9. AH (t) = U −1 (t.

qj ). qj )]) .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. Combining the two terms gives qj+1 . use the Campbell-Baker-Hausdorﬀ formula. . . . These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|. t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . [A. eA eB = eC with C = A+B+ (9.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . t′ |q. ˙ 2π 1 1 1 [A.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 ′ . which is a hamiltonian function in which the operators are replaced by real-numbered variables. 2 12 12 (9. Then qj+1 . with the correction1 being of order (∆τ )2 . B]. (9. n |q (9. B] + [A.q . The hamiltonian is an operator H = H(P. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . Q) = K(P ) + V (Q) = (P 2 /2M ) + V (Q) one can split e−i H∆τ ≈ e−i K∆τ e−i V ∆τ .12) where the transformation between coordinate and momentum space involves q|p = ei p.. tj+1 |qj . H(P. 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 .. t = q ′ |e−i H(t −t) |q . Q) expressed in terms of the operators P and Q. 2π (9. tj+1 |qj .8) Choose t as the starting point with |q = |q.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . B]] + [[A.16) this. 2π (9.14) .10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. QS |q ′ ≡ q ′ |q ′ . |q1 q1 |e−i H∆τ |q .11) (9. tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . B] + .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . ′ (9. . (9. t′ |q.15) By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj .

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

2 (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 ≡ 1.22) = 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). Dα F [α] = Dα F [α + β].24) (9. A useful property of functional integration is the translation invariance.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions.25) (see Exercises). Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.23) Having deﬁned the Gaussian integral. 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. 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.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. (9. Consider the gaussian functional as an example. y) = ∗ m.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.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. .27) (9.21) again a multidimensional integral. (9. α(x) = n αn fn and K(x. the following integrals can be derived for a real symmetric or complex hermitean kernel K.

(9.g.33) (9.35) For applications to fermion ﬁelds. α(y) = δ(x − y). δα(x) (9. δα(x) or in discretized form δ 1 ∂ ∂ with . y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. αn = δmn . dx dy θ∗ (x) K(x.e. products of the same pair appear. n θn fn (x) dx θ∗ (x) θ(x) . Examples of functional diﬀerentiation are (9. e. (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.29) (9. y) θ(y) . i. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . however. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. the Gaussian-type functionals.31) δ δ α = dy α(y) = dy δ(x − y) = 1. δα(x) δα(x) δ αKβ = dy K(x. F [θ.38) . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). (9. θ2 = 0.32) (9. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . In the expansion of the exponential (from second to third line).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. αy = δxy .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.30) (9. In principle the deﬁnitions of functionals is the same.34) (9. θη = −ηθ. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. 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

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

The importance of Z[J] is that the time ordered product of operators can be obtained from it.70) for the generating functional is in fact ill-deﬁned. . t′ |q. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. δJ(tn ) hence the name generating functional. T → i ∞ and T ′ → −i ∞. .6 Euclidean formulation The above expression (Eq. . t) (9. Since we have (neglecting multiplicative factors). T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. = (i)n 0|T Q(t1 ) .69) The factor that has been divided out in the ﬁrst line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ). T = φ0 (q. (9.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. .70) one immediately ﬁnds δ n Z[J] δJ(t1 ) .2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. ˜ t′ → ∞. thus T →i∞ lim q. J (9.71) 9. t 0 0out |0in J . The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. 9. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. Q(tn )|0 . J=0 (9. . Better is the use ˜ of imaginary time t = −it. t′ ) q ′ . T ′ |Q. dq dτ + J(τ )q(τ ) .67) = = φ0 (q.68) (9. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . t). t|Q.

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

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

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

5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . . d4 xn G(n) (x1 . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . J(xn ). (10.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . 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. . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . . As discussed in the previous section (exercise 9. . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).10) J=0 In order to determine ∆F .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10.3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .e. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. xn ) J(x1 ) . .6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). with Z0 [0] = 1. . (2π)4 (k 0 + iǫ)2 − E 2 (10.11) Depending on the path choosen in the complex k 0 plane (see exercise 7. .9) i. . (10. . 2 2 → (10. xn ) = x2 xn . Furthermore.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. . .13) . .

17) d4 k e−i k·x . The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . the advanced Green’s function ∆A (x) = satisfying ∆A (x. (10.16) C where C is a closed contour in the complex k 0 -plane. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . 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. ∆(x) = − d4 k e−i k·x .15) e−i k·x d4 k . 10.4). (10. (2π)4 (k 0 − iǫ)2 − E 2 (10. t) = 0 for t > 0. ∆F = ∆C . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).18) Thus we see various solutions. It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x). The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).14) They are solutions of the inhomogeneous equation. i. µ 1 1 • Firstly. . φ+ (0)] = (2π)3 2E are also shown in exercise (7. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.22) (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 . t) = 0 for t < 0.20) = exp − 2 where or (see ﬁg. 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).Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .e. (2π)4 k 2 − M 2 (10. 4 k 2 − M 2 + iǫ (2π) (10. (2π)4 k 2 − M 2 (10.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x .

we have G0 (x1 . because this is the only quantity entering Z0 [J]. x3 . 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. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . x2 . or diagrammatically (4) (10. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) .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 10.3) 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). the same result as above. We ﬁnd (see also section 7. = θ(x0 − y 0 ) 0|[φ+ (x).24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . . φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility.25) x1 x2 x3 = x4 + + .

34) Introducing a vertex point to which four propagators are connected.33) (4) i −i d4 z LI g 4! d4 z 1 δ i δJ(z) 1 exp − i 2 J ∆F J 2 −3 ∆2 (0) + 6i ∆F (0) F + d4 x ∆F (z − x) J(x) 4 d4 x ∆F (z − x) J(x) 1 exp − i 2 J ∆F J . ≡ −i g d4 z (10. x3 . x4 ) = G0 (x1 .Feynman diagrams for scattering amplitudes 89 10. x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. x3 .29) This expression will be the one from which Feynman rules will be derived. the generating functional can be written.g ) (x1 .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. (10.35) z .g ) (x1 . x2 ) = i ∆F (x1 − x2 ) = G(4. (10. x2 ) = G0 (x1 . Z[J] = = Dφ exp i exp i d4 z LI (10. (10.g 0 ) = G0 = 1.31) G(2. i.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 .32) (10. x2 .30) and in zeroth order G(0. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].28) 1 exp − i 2 = exp 1 2 d4 x d4 y exp i δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)] φ=0 . L (φ) = L0 (φ) + LI (φ). (2) (0) (10.27) When interactions are present. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar.

. 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]. .40) in−1 n! d4 x1 . did not contain parts that could be written as products of simpler Green’s functions. . d4 xn G(n) (x1 . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . (10. . . . (10.g 1 ) = 1 8 z 1 8 . . Z[J] = n in n! d4 x1 . . . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 .37) G(2. which also was the only Green’s function for which the diagram was connected. . . xn ) J(x1 ) . Connected Green’s functions We have now seen the Green’s functions as the quantities appearing in the expansion of the generating functional Z[J]. d4 xn G(n) (x1 . i.36) J J from which one obtains G(0.39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. (10. .g ) (x1 .38) The result for the 4-points Green’s function is left as an exercise. . J(xn ). . xn ) J(x1 ) . In the free case we have seen that the exponent of Z[J] contained all essential information. . . (10. . δJ(xn ) . . . . (10. x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . G(n) (x1 .41) in terms of socalled connected Green’s functions. Diagrammatically this exponent only contained the two-point function.Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. J=0 (10. .e.42) x1 G(n) (x1 . . This remains also true for the interacting case. xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . . . . J(xn ) c (10.

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

47) . the vacuum expectation value of the ﬁeld φ(x) is zero. out . e. . thus β.g.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 ). φout and the interpolating ﬁeld φ(x). x4 10.e. out = |p′ .3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. in = |α. out|α.Feynman diagrams for scattering amplitudes 92 i. x2 . (10.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. . out| and S † |α. in = S|α. in ⇔ β. (10. in|S = 0. one ﬁnds for the connected 4-point Green’s function at order g. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . Proof: α. For the free scalar theory. in|S|p′ . in = p. . in| (choose ϕ0 = 0). translation invariance). pn . in = β. out = δαβ ↔ SS † = 1. in = |p1 . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . in|S = β. x4 ) = c 1 x3 z (10. out|α. Next. out (suppressing except momenta all other 1 m quantum numbers). in = p. Proof: 0. implying the absence of tadpoles. considered explicitly. . Sβα ≡ β.45) x2 as the only surviving diagram (see exercises). in|SS † |α. (10. x1 G(4. out|p′ . p′ . p. then (n) (n) Gsc = Gc for n ≤ 3. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). in|S = α. (2) The one-particle state is invariant (conservation of energy and momentum. in|p′ . t = −∞ φin (x) t = +∞ φout (x) φ(x) where φin and φout transform under the Poincar´ group as scalar ﬁelds and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . ∂µ ∂ µ + M 2 φin (x) = 0. this will be translated to the action on ﬁelds. . in into ﬁnal state particle states (out-states) |β. out = p|p′ . . (2) (10. out| = eiϕ0 0.g ) (x1 . out|p′ . x3 . out| ⇔ |α.44) J=0 For the interacting theory. in = p. out .

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

φ(y)] f (y) e e .63) .60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x).56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. 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). express it in terms of advanced and retarded Green’s functions. e e ] = [A. δJ(z) (10. φ(x). B + C]e e . 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.57) i [φin (x). y) δ δJ(y) : F [J]. (10. t i Ut∞ [J] φ(x)U−∞ [J] (10.61) 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. one has [A.62) where F [J] is some (arbitrary) functional. eB ] = [A. z) . δJ(z) (10. B] and [A. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. z) δU [J] . φ− f (10. [A. φ(x). B] (for the case that [A. we can. provided [A. C] are c-numbers. just as we did for φ(x). which is equal to the expression e in which the creation operators are placed left of annihilation operators. (10. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. B] is a c-number).58) In order to ﬁnd a solution to this equation note that.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. S U [J]] = √ Z δ S U [J]. δJ(z) Taking the diﬀerence between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. e = i.59) (10. B]eB . z) d4 y d4 z ∆(x − y)K(y. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10.e.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).

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

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

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

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

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

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

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

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

the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ). q).e. . Its square root. It is a useful quantity. s is for obvious reasons known as the center of mass (CM) energy. It is determined by the energy-momentum four vector p = (E. . 103 . The ﬁrst is the CM system.7) pb = cm (Eb .5) √ is referred to as the invariant mass squared. . p1 . p) which satisﬁes p2 = E 2 −p2 = m2 . In that case cm pa = (Ea . −q). . (11. (11. . .1 kinematics in scattering processes Phase space The 1-particle state is denoted |p . . . pn ). a a b b the quantity s = P 2 = (pa + pb )2 . . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . A physical state has positive energy.6) (11. pb we start with the four vectors pa = (Ea . i n d3 pi . To be speciﬁc let us consider two frequently used frames. (2π)3 2Ei i=1 (11.3) which is useful because the total 4-momentum of the ﬁnal state usually is ﬁxed by overall momentum conservation. + n. (11. . . s = (p1 + . Here s is the invariant mass of the n-particle system. pa ) and pb = (Eb . . pb ) satisfying p2 = m2 and p2 = m2 .1) = (2π)3 2E (2π)4 (proven in Chapter 2). . and the total momentum four-vector P = pa + pb . This is generalized to the multi-particle phase space dR(p1 . pn ) = and the reduced phase space element by dR(s.Chapter 11 Scattering theory 11. . + pn )2 .4) For two particle states |pa . (11. i. In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . for instance for determining the threshold energy for the production of a ﬁnal state 1 + 2 + . . The phase space is determined by the weight factors assigned to each state in the summation or integration over states. . For two particles.2) (11.

m2 .10) The function λ(s.16) . (11. m2 ) = 4(pa · pb )2 − 4p2 p2 .13) (11.15) Explicit calculation of the reduced two-body phase space element gives dR(s. m2 .11) (11. 2 s s − m2 + m2 a b √ . which in the speciﬁc case a b also can be expressed as λ(s. p1 . m2 ) a b . 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = . m2 . Also in this case one can express the energy and momentum in the invariants. m2 . 0). 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. 2mb λ(s. 4π s 4π 8π s 4π (11. 4s (11.9) (11. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. p2 ) = = where λ12 denotes λ(s. mb and s). 2mb (11. = 2 s λ(s. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . ptrf ). a (11. p1 . In that case trf pa = (Ea . Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . m2 ) a b .8) (11. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . for instance.12) pb = (mb . m2 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 ) is a function symmetric in its three arguments. 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.14) One can. m2 ).

11. The variable s is always larger than a c b d the minimal value (ma + mb )2 . (11. . in exclusive measurements all particles are detected. 2s (11.25) dR(s. Consider the process a + b → c + d. H1 + H2 → h + X. = (11.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. py ). Consider the 1-particle inclusive case.17) (11. In inclusive measurements no particles are detected in the ﬁnal state. Examples appear in → π− + p (11. usually is a 2-particle state. there is usually a preferred direction.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. 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 . in which one particle is detected.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .18) (11. E . however. usually many particles are produced. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. Instead of the scattering angle.21) (11. The various possibilities are referred to as diﬀerent reaction channels. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11.4). tmax min λcd /4s.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. 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 . pc .27) with m2 = m2 − p2 = m2 + p2 + p2 . A speciﬁc reaction channel starts contributing at the threshold value (Eq. pd ) = 4π s 4π 8π s 4π dt dt √ √ . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .19) (11. Of course there are not only 2-particle channels.. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). pT = (px . The initial state. An often used set of invariants are the Mandelstam variables. Writing E = mT cosh y. At high energies..20) → π0 + n → π+ + π− + n → . for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.22) (11. pz = mT sinh y.

t. t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. This is known as crossing symmetry. tu = s. t = (p1 − p3 )2 = (p2 − p4 )2 . tt = s. ¯ tt = (pa − p¯)2 = s.30) (11. ¯ 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. η = − ln(tan(θ/2)) = tanh−1 (cos θ).Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. ut = u) = Mab→cd (s. u).31) . c bd Mad→c¯(su = u. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). t. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. u = (p1 − p4 )2 = (p2 − p3 )2 . For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. (11. b ut = (pa − pd )2 = u. u) with s = (p1 + p2 )2 = (p3 + p4 )2 .2 Crossing symmetry In the previous chapter. t-channel and u-channel processes respectively. (11. 11. uu = s) = Mab→cd (s. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. which means that rapidity distributions dN/dy maintain their shape.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1.28) which (only involving angles) is easier to determine. uu = (pa − p¯)2 = s. ¯ b (11. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. u).

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

For the (proper) decay width one thus has 1 N Γ0 = . . For this we employ in essence Fermi’s Golden rule stating that when the S-matrix element is written as Sf i = δf i − (2π)4 δ 4 (Pi − Pf ) iMf i (11. dt (11. .35) This quantity is not covariant. T · V ρa (11.34) i. pn ).36) (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.Scattering theory 108 Although this at ﬁrst sight does not look covariant. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . (11. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. 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 .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. 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 . the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. N and T · V are covariant.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. .e. (11.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves).T V.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). it is. N 1 . . the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . σ= √ 2 λab T · V (11.38) (in which we can calculate −iMf i using Feynman diagrams). . as expected.

(11.42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. (11. q n ). 16π 2 s (11. θ) = −8π s ℓ (in analogy with the expansion for f (E. see 11. implies for the scattering matrix M .46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. θ). The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. 8π s 2π 8π s .47) M (s. 11. i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . (11.43) dΩcm = π M (s. p1 . ˆ ˆ Inserted in the unitarity condition for M . . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . q n ) q n qn √ Mni (q i . pn ). .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 . .41) (11. The result is N = |Mf i |2 dR(s. t) λab 4π dt. respectively. . . p1 . p1 . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. i. . .e. . . for which one has dσ/dΩ = |f (E. . note the sign and cos θ = qi · qf ).25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . . for which the √ amplitudes squared have been calculated in the previous chapter. pn ) |M |2 q 32π 2 m2 dΩ |M |2 .Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). pn ). θ)|2 . √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). 2 λab and for instance for two particles dσ = q ′ M (s.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. θ) or M (q i . (11. θcm ) √ q 8π s 2 2 (11. q f ). θ) in quantum mechanics. (11.

Using σ= dΩ q ′ M (s. q ˆ i. 2i q 2i q (11. (11.54) ℓ . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). (11. q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni .49) or Im Mℓ = q |Mℓ |2 . ℓ (11.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . the total cross section can be determined. In general a given channel has |Sℓ (s)| ≤ 1.50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift.e.52) From the imaginary part of M (s. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . (11.53) The diﬀerence is the inelastic cross section. n (11. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . (11.51) and for the case that this is the only channel (purely elastic scattering. 0). θcm ) √ q 8π s 2 . parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel.

the time-evolution operator.e. The quantity Γ is precisely the width for unstable particles. pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. 0). ∝ U (t. . . which is given by a sum of the partial widths into the diﬀerent channels. We note that unitarity is generally broken in a ﬁnite order calculation. η = 0.e. 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. For an unstable particle one expects that U (t. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior.55) (note that we have disregarded spin).Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. i. such that |U (t. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . To check unitarity we need the amplitude at order e4 .56) (again disregarding spin). however. 11. The product of amplitudes is of order e4 . 0)|2 = e−Γt . 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. For instance the tree-level calculation of Feynman diagrams at order e2 was real. (q Mℓ )ij (s) = −M Γi Γj . . in which case. For the amplitude in a scattering process going through a resonance. s − M 2 + iM Γ (11.57) This shows that Γ is the width of the resonance. (11. but still real. speciﬁcally one has for t > 0 that the Fourier transform is dk e 0 −ik0 t ∆(k) ∝ e−ik t dk (k 0 − Ek + iǫ)(k 0 + Ek − iǫ) 0 0 (t>0) ∝ e−iEk t . i. .5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. 0) ∝ e−i(E−iΓ/2)t . 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 . σel = σinel . it is straightforward to write down the partial wave amplitude.58) . This is what we will do to discuss decay widths. The prescription for the pole structure.

88 MeV. which is attributed to neutrinos. dt s(s − 4m2 ) with f (t. This is a narrow resonance discovered in 1974. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . • The Z 0 resonance in e+ e− scattering with M = 91. Its mass is M = 3096. u) = g(t. i. • The J/ψ resonance in e+ e− scattering.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. 2 q (s − M 2 )2 + M 2 Γ2 (11.2 GeV. Three neutrinos explain the resonance shape. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . Exercises Exercise 11. The cross section at resonance is about 30 nb.Scattering theory 112 (Prove this using the unitarity condition for partial waves). Note that because of the presence of a background the phase shift at resonance may actually be shifted. At resonance the cross section σT (π + p) is about 210 mb. its width Γ = 120 MeV.4) can be written as 4πα2 dσ = {f (t. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. Knowing that each neutrino contributes about 160 MeV (see next chapter). From this one sees that a resonance has the same shape in all channels but diﬀerent strength.49 GeV. The half-width of the resonance is M Γ. t) + g(u. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . a fast change in the phase shift for a narrow resonance. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. u) + f (u. where furthermore σinel = 0. Γ = 2. the partial width into e+ e− is Γee = 5. 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). The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb.59) reaching the unitarity limit for s = M 2 . Its mass is M = 1232 MeV. Limiting ourselves to a resonance in one channel. the full width is Γ = 88 keV.26 keV. All these decays can be seen and leave an ’invisible’ width of 498 MeV.1 Show that the cross section for electron-electron scattering (exercise 10. M2 − s (11. This characteristic shape of a resonance is called the Breit-Wigner shape. t)} . u) + g(t.e.

999 877 Γ and Γ(π − → e− νe ) = 0. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . As discussed for unstable particles. Exercise 11.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0.8.2).000 123 Γ. Thus.3 Complete the calculation of the decay width and lifetime of the pion by including the appropriate phase space factors and using the calculation of the squared matrix element in Exc.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. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude.Scattering theory 113 Exercise 11.2 Calculate the ﬂux factor. cm |va − vb | in the center of mass system for two colliding particles a and b. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. ρcm ρcm a b 1 cm . Calculate now fπ . 4. writing N (s) M1 (s) = 2 + iM Γ . Exercise 11. Compare the results with the experimental π − lifetime τ = 2.

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

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

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

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

inf

(12.3)

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

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

The standard model

115

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

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

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

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

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

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

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

**A geometric picture of gauge theories
**

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

1 The

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

λ7 =

i

−i

1 1 λ8 = √ 3

−2

x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). However.16. but expresses them with respect to diﬀerent bases. should be independent of such transformations.e. (12.19) This gives rise to a (path dependent) phase in each point. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x).16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). (12. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).15) ψ(x). = . i. In principle such a phase in a given point is not observable. i. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). 12. ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). -dx -dy µ µ dyµ µ dx For a vector. 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.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. which connects two identical vectors.The standard model the basis for x G and x+dx 117 G. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). The relation in Eq. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). ds which is solved by dψ(s) ds = dxµ .14) (12. (12. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed.e. (12.

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

The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. for the vacuum Qa |0 = 0. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0.e. . while the lagrangian including the nonzero groundstate expectation value chosen by nature. therefore a choice must be made. It is only invariant under rotations around the 3-axis. The classical groundstate appears degenerate.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. (12. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. 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. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . . φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0.23) F +η the latter only forming a minimum for m2 < 0.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. ϕ2 and η. i. They are known as the Weyl mode or the Goldstone mode: Weyl mode. . The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. The situation now is the following.) terms constitute interaction terms.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . In this case the spectrum is described .25) 2 2 2 Therefore there are 2 massless scalar particles.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). In this situation one speaks of spontaneous symmetry breaking. The η massless particles are called Goldstone bosons. 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 . etc. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. In the case of degeneracy. has less symmetry. while the quadratic terms must correspond with non-negative masses. (12. say 0 0 . In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . Any constant ﬁeld with ’length’ F is a possible groundstate. ϕc = 0|φ |0 = (12. (12.

This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. 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. there must be a nonzero expectation value 0|Jµ (x)|π (k) . a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . and one must have mπa = 0. T = τ /2.. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). a and a′ are degenerate states.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. . M = . π (12. This means that they are operators that create states from the vacuum. (12. Taking the derivative. Using the Dirac equation. . the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. The generators Qa (in that case the rotation operators Lz . i. 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′ . π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. V µ = ψγ µ T ψ. if the generators Qa generate a symmetry. ψ −→ ei α·T ψ. ψu mu md ψ = ψd . Goldstone mode.e. i. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). strange. (12. Neglecting at this point the local color symmetry. . Including a sum over the diﬀerent ﬂavors (up.e. irrespective of the fact if they annihilate the vacuum. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. T ] ψ.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. [Qa . ∂ ∂ (12.31) . .29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). etc. called ﬂavors. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). denoted |π a (k) .26) for all the states labeled by a corresponding to ’broken’ generators. it is easy to see that one gets ∂µ V µ = i ψ [M.) one can write L = f ψ f (i/ − Mf )ψf . Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab (fπ = 0) (12. a ∂ µ Jµ = 0. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . H] = 0.e. a massless Goldstone boson for each ’broken’ generator.32) (12. down. = ψ(i/ − M )ψ.The standard model 120 by degenerate representations of the symmetry group. A bit more formal. (12. i.

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

42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). Initially there are 3 massless gauge ﬁelds (each. thus 9 independent degrees of freedom.39) (12. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. One may wonder about the degrees of freedom. (12. having two independent spin components) and three scalar ﬁelds (one degree of freedom each). but one has 1 massless gauge ﬁeld (2). 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ).38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν .37) (12. 0 ϕc = 0|φ |0 = 0 . We have the possibility to perform local gauge transformations. The symmetry is broken in the same way as before and the same choice for the vacuum. i. 0 φ= (12. After symmetry breaking the same number (as expected) comes out.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. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). The diﬀerence comes when we reparametrize the ﬁeld φ. . Dµ φ = Gµν = ∂µ φ + g Wµ × φ. again 9 degrees of freedom. . is made. while the number of massive gauge bosons coincides with the number of ’broken’ generators. 0 0 0 = . . 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 + . 4 2 where Since the explicit (adjoint. (12. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m .40) a Dµ φ = ∂µ φ − ig Wµ La 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).e.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 . like a photon. (12. in this case three-dimensional) representation reads (La )ij = −i ǫaij . Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. made into a gauge theory. .The standard model 122 12.

). Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. i. U(1)Y U(1)Y (12. muon µ− or tau τ − ). (12. that gives a good description of the physical world.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. τ ).) and the gauge bosons responsible for the strong. 2 2 = ∂µ R + i g ′ Bµ R. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. its corresponding neutrino (νe .44) which has an SU (2)W symmetry under transformations eiα·TW .45) (12. left. The procedure. explicitly L −→ e−i β/2 L. R −→ e−i β R. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. ψR/L = 2 (1 ± γ5 )ψ decouple. 2 (12.48) (12. = ∂µ L − (12. electromagnetic and weak forces.e. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. while the lefthanded particles form a doublet. 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. νµ and ντ ) and their antiparticles. neutrinos. R. etc. down. This is written as 3 Q = TW + YW .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. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.The standard model 123 12.47) and YW is considered as an operator that generates a U (1)Y symmetry. the quarks (up. muons. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 .51) . etc. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . (12. the leptons (elektrons.and righthanded particles. 1 As they are massless. ∂ ∂ ∂ (12.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. however. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.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.

L where L (h1) (h) =L (h1) +L (h2) .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1.56) 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. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). h) (12.58) (12. (12. (12. −V (φ) L and (h2) = −Ge (LφR + Rφ L).59) . . . . + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .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 . (12.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. 4 4 In order to break the symmetry to the symmetry of the physical world. the U (1)Q symmetry (generated by the charge operator). For the classical ﬁeld the choice θ4 = v is made. Using local gauge invariance θi for i = 1. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . . this leads to the lagrangian L (h1) . Putting this in leads to L (f ) =L (f 1) +L (f 2) . † i i g Wµ · τ − g ′ Bµ )φ. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. leading to the parametrization 1 0 φ(x) = √ (12.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .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. 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. 2θ 2θ 4 cos W cos W 0. g ′ /g = tan θW . (12.65) The weak mixing angle is related to the ratio of coupling constants. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ . . 2 Zµ (12. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.66) From the coupling to the photon. 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) (12.67) The coupling of electrons or muons to their respective neutrinos. 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 + .63) (12.69) . 4 2 MW g2 v2 = .60) (12. 2 (12. we can read oﬀ e = g sin θW = g ′ cos θW .61) (12.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ . The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ .

With the choosen vacuum expectation value for the Higgs ﬁeld. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).231 2.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.74) (12. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.e. 2 2 2 (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. 3 = TW .e. giving rise to a running coupling constant after renormalization of the ﬁeld theory. one has the relation g2 e2 1 GF √ = .73) (12. g ′ and v determine a number of experimentally measurable quantities.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. g2 MZ 2 2 (CV + CA ). as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). i.166 4 × 10 0. 80. Finally we want to say something about the weak properties of the quarks. 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. (12.g. 91. The coupling to the Higgs particle is weak as the value for v calculated e.77) (12. .80) First.40 GeV. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. v (12. (12. Comparing this with the total width into (invisible!) channels. (12.19 GeV.71) GeV −2 = = = = . −5 (12.75) 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 . from the MW mass is about 250 GeV. (12. 1.5 MeV (exp. i. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . muon or tau.72) (12.76) with CV CA 3 = TW − 2 sin2 θW Q. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. me /v is extremely small.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.

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

UR = uR cR tR . u u where we include now the three lefthanded quark doublets in QL . . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . the weak mass eigenstates are L (h2.34. (12.82) space.227. each of these containing the three families. ms and mb ). d d (12.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. 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 .86) the unitary CKM-matrix introduced above in an ad hoc way.22 and η ≈ 0.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .e. (12. A ≈ 0. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . magnitudes of the entries in the CKM matrix are nicely represented parametrization 1 − 1 λ2 λ λ3 A(ρ − i η) 2 1 2 −λ 1− 2λ λ2 A V = 3 2 λ A(1 − ρ − i η) −λ A 1 12.82 and ρ ≈ 0. e. † Λu Λ† = Vu G2 Vu . i.83) where Vu and Vd are unitary matrices. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues.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. 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 .g.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. respectively). The Higgs ﬁeld is still limited to one complex doublet. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .

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 . 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 . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . As before. (12. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn 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.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. since the weak current is the only place where they show up. (12. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . mµ and mτ . i. EL = Ve† EL . (12. In this case.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . EL = Ve EL .ℓ) = −LφΛe ER − ER Λ† φ† L. e N L= L EL (12.90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER .5×10−3 eV2 .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. there is no family mixing for massless neutrinos. one can write Λe = Ve Ge We and we ﬁnd † L (h2. we ﬁnd massless neutrinos.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. The mass ﬁelds ER † mass † mass = We ER . and obtain L (h2. decoupling from all known interactions.e. (12.92) . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL .89) with three righthanded neutrinos added to the previous case. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .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 . UPMNS a parametrization that in principle also could have been used for quarks.

one being massive. coupled by a Dirac mass term.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. For the neutrino sector.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . 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.99) . We use here the primes starting with the weak eigenstates. Thus (disregarding family structure) one has two Majorana neutrinos.95) 2 In order to have more than a completely decoupled sector. one must for the neutrinos as well as charged leptons. The mass matrix can be written as ML |MD | eiφ M = (12.ν =− containing three (CP-violating) phases (α1 . For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. 1 c ∗ c L mass. For these light Majorana neutrinos one has.and righthanded species then just form a Dirac fermion. however. 1 ∗ c c (12. α2 and δ). Absorption of phases in the states is not possible for Majorana neutrinos. are equivalent to two decoupled Majorana neutrinos (see below). The way to circumvent this is to introduce as in the previous section righthanded neutrinos. This is called the see-saw mechanism (outlined below). couple the right. For the leptons. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors.97) As the most general mass term in the lagrangian density we have L mass = − = which for MD = 0 is a pure Majorana lagrangian and for ML = MR = 0 and real MD represents the Dirac case. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ).94) M L NL NL + M L NL NL . the massless and massive Majorana neutrinos. Actually.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos.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.ν = − MR NR NR + MR NR NR .98) (12. hence the mixing matrix becomes iα /2 0 0 e 1 (12. the left. is to add in Eq. with for the righthanded sector a mass term MR . a unitary matrix relating them to the weak eigenstates. Thus 2 Υ ′ = nc + nL . as above.c.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. (12. (12. 12. 2 although this option is not attractive as it violates the electroweak symmetry. − MD MR nR 2 (12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. 1 L Υ ′ = nR + nc . 2 R ψ = nR + nL . 0 0 1 L mass. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.

For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . H] = 0.e. Prove that if the generators Qa generate a symmetry. The diagonalization is analogous to what was done for the Λ-matrices and one ﬁnds U M U T = M0 with a (unitary) matrix U . (12. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .The standard model 131 taking ML and MR real and non-negative. Exercises Exercise 12. 2 with real masses Mi . related via c Υ1R = U nL .101) Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2 . T ] ψ. i ψ {M.1 Consider the case of the Weyl mode for symmetries. Exercise 12.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . 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). This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. .3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 . (12. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. T } γ5 ψ. a and a′ are degenerate states.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)) .103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .2 Derive for the vector and axial vector currents. L Υ2L nR Υ1 and Υ2 .e. i. i. [Qa .105) for i = 1. (12. Υ2R nR (12. Exercise 12. 2 U (M M † ) U † = M0 .

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

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

Sign up to vote on this title

UsefulNot usefulrelativistic quantum mechanics

relativistic quantum mechanics

- Quantum Field Theory 2013
- Introduction to quantum field theory, based on the book by Peskin and Schroeder
- An Introduction to Quantum Field Theory - Peskin
- Huang K. - Quantum Field Theory
- Srednicki M. Quantum Field Theory - Solutions Manual(Web Draft, 2006)(156s)_PQtb
- PESKIN - Quantum Field Theory
- An Introduction to Quantum Field Theory(Peskin)
- Srednicki Ms Qft Solutions Rev
- Weinberg, Steven - The Quantum Theory of Fields Volume II
- Lahiri and Pal-A-First Book of Quantum Field Theory-2nd Ed
- Math Vectors v2
- Gate Calculus
- Computational Fluid Dynamics
- appriviation 8.pdf
- What is a Vector
- Rigid transformations
- AE264_Spring2014_HW1
- Spherical Base Vectors
- 9th_physics_CHAPTER_4_S.Q_numerical_problems.pdf
- Lecture_Notes__2_1_-_2_3
- Balancing
- Balance 2
- Einstein Derivation
- Additional Mathematics Project SPM
- Weinberg the Quantum Theory of Fields Vol 1 Foundations
- 1.1intro to Physics
- Vectors
- AMC110S Statics Mod0-Introduction.ppt
- Geo and Discret
- h65_Vectors_in_R2_and_R3.pdf
- QFT-0