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

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

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

3 2.2 36 22 .2.4 6. 1994.1 3. 1995. Quantum Field Theory.1 2 2 3. Corresponding chapters in books of Ryder.3 2. Vol.1 5.4 3. Srednicki.3 3. 4.2 3.3 6. I: Foundations.1 1.5 5. The notes contain all essential information.3 References As most directly related books to these notes. 1980. M.3. 6.3 2.2 6.7 2. 1996. 2. II: Modern Applications. 3. M.2 3. 1985. 1.1 2.2 3. 2.1 6. Diagrammatica.3 3. The quantum theory of ﬁelds. but are rather compact. S.3 2.2 2. M.1. L.5 4. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012).2 1. Peskin & Schroeder and Srednicki.E. Veltman. These notes Section 1. C. 1995. Cambridge University Press.2 6.3 3. Schroeder. Quantum Field Theory.3 4. 5.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.H.1 3. Quantum Field Theory.4 3. Cambridge University Press.3 3.4 2. Cambridge University Press. I refer to the book of Srednicki [1] and Ryder [2]. Zuber. Itzykson and J. Vol.V. 2007.2 3.1 2.2 4. Ryder. Cambridge University Press. McGraw-Hill.2 3.4 4.2 2.8 3. 3.4 2.1 4. 3.-B. Peskin and D.3 2. Addison-Wesly.3 3.2 2. Cambridge University Press. Weinberg. Other text books of Quantum Field Theory that are useful are given in refs [3-6].4 3. An introduction to Quantum Field Theory.5 2.5 2.

4 10.2 12.3 8.6 4.1 11.7.3.3 3.7 4.1. 6.4 8.5 4.1 20.3 7. 2.1.2 10.5 4. 4.1 12. 2.1 9.1 20.1 10.4. 20. 4.4 12.2.2 8.5 .3 11.1 9.7 5.2 5.4 2.2 8.1 7.2 9.1 8. 6.6 3.4.5 7.3 6 8 9 Srednicki 3 3 3 37 5. 9.3 2.4 12.6 9.2 11.5 7.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.3 6.8 5 4.2.4 7.4 These notes Section 7.5 11.10 8.3 9.3 7.5.5 2.3. 4. 6. 6.2.6 6.6 8.2 7. 9.6 3.3 8.5 9.1 5.4 3.3 10.3.6 10.8 6.4 4.5 9.3 12.6 3.4 9.2.1 20.5 6. 8.1. 4.4.2 4.3 20.5 6 7 6 6. 6.

2 .3 • First two sentences of Chapter 5 have been modiﬁed.5 Corrections 2011/2012) • Extension of Exercise 3. • Slight rewriting of discussion of Majorana mass term in section 6.

˜ ψ(p) = i d3 r exp − p · r ψ(r). Corresponding position and momentum operators do not commute. (1. pj ] = i δij . In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). S ≫ .582 119 15 (56) × 10−22 MeV s.4) where δij is the Kronecker δ function. 1 (1. In special relativity a special role is played by the velocity of light c.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. usually given divided by 2π.054 571 68 (18) × 10−34 J s = 6. It is given by pop ψ(r) = −i ∇ψ(r).5) One can also choose a representation in which the momenta of the particles are the dynamical variables.2) In the limit that v ≪ c one reaches the non-relativistic domain. being eigenvalues of the position operators. the action of the momentum operator in the coordinate representation is not as simple as the position operator. (1.Chapter 1 Introduction 1.6) . c = 299 792 458 m s−1 . One can talk about states |r and the wave function ψ(r) = r||ψ . In quantum mechanics. the position can in principle be determined at any time with any accuracy. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.1) In the limit that the action S is much larger than . Indeed. They satisfy the well-known (canonical) operator commutation relations [ri . (1. (1. ≡ h/2π = 1.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. In quantum mechanics a special role is played by Planck’s constant h. quantum eﬀects do not play a role anymore and one is in the classical domain. (1. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-deﬁned concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions.

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.

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

Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. . but more important has the same eﬀect on lefthandside and righthandside. 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. Each of these permutations leads to a minus sign. (1. for instance. (1. Note that in this equation one has on left.33) (1. a∞ = 4πǫ0 2 /me e2 . p). 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.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). ∂t (1.∇ . ∂3 ) = ∂ ∂ ∂ ∂ . p) = (E.34) (1. ∂2 . = ∂ . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 .36) Exercises Exercise 1. ∂1 . The length squared of ∂ is the d’Alembertian operator.28) ν ν Note that gµ with one upper and lower index.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . constructed via the metric tensor itself. It is an invariant tensor. (1. not aﬀected by Lorentz transformations. because one has (1. where m is called the mass of the system.27) It is easy to convince oneself of the nature of the indices in the above equation. = −24. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. .and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector.35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . (1. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . (1. starting from ǫ0123 = 1.32) .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. . deﬁned by ∂2 − ∇2 . explicitly written as (p0 . 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 . ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . g0 = g1 = g2 = g3 = 1. ′ ′ .

1. Also for the coordinates [a− . This demonstrates how a careful use of units can save a lot of work. 1. 0. (d) Use the value of the gravitational constant GN / c5 = 6. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. n2 . a1 . ˆ ˆ ˆ ˆ These are n1 = (0. (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). by a simple few-line reasoning [For instance: If {µ. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . (b) The coordinates (a0 . ˆ ˆ . but only appropriate combinations.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . ]. Compare it with the proton mass and use Eq. . . a+ . 0. How are a+ and a− related to a+ and a− . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). . Exercise 1. a1 .(A · B) C using the properties of the tensor ǫijk given in section 1. 2.13 to give its actual value in kg. α and me c2 = 0. 1) ˆ . e.2 Prove the identity A × (B × C) = (A · C) B . n1 . 0) ˆ n3 = (0.511 MeV. which is the Compton wavelength for the Planck mass. Exercise 1. ǫ0 . σ} is a permutation of {0.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). a2 ] we can ﬁnd basis vectors n− . re = e2 /4πǫ0 me c2 to the Compton wavelength. [Note: what is the MKS unit for V/T?] Exercise 1.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a1 . g−− . 0) ˆ n0 = (1. a+ . 0. n1 . a2 . me . One does not need to know . n3 . 0) ˆ n2 = (0. n+ . Also construct and calculate the Planck length Lpl . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . g+− and g−+ . ν. 1. 0. 0. n2 . a+ . a3 ) are the expansion coeﬃcients using the basis vectors n0 . 1. ρ. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . c. 3} the index α can only be equal to one of the indices in ǫµνρσ .3. 0. a2 ] or [a− .

(2.2) Acting on the wave function one ﬁnds for a free particle. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. namely plane waves characterized by a wave number k. 2M (2.1 The Klein-Gordon equation In this chapter. φk (r. ∂t2 (2. 8 . t) = exp(−i k 0 t + i k · r).6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. ∂t p −→ pop = −i∇. But the replacement 2.3 are not covariant.Chapter 2 Relativistic wave equations 2.3) Equations 2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). 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. 2 + M 2 φ(r. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. Although it is straightforward to ﬁnd the solutions of this equation. t) = − ψ(r.2.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . (2. ∂t 2M (2. p2 = pµ pµ = E 2 − p2 = M 2 .4) where M is the particle mass.7) k 0 = ± k2 + M 2 = ±Ek . t) = 0. p). 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. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.5) with (k 0 )2 = k2 + M 2 . (2. t) = ∂2 − ∇2 + M 2 φ(r.1 and 2. one obtains (2. E = p2 . t).

φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek .12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. ∂µ j µ = 0. Then. for which we need the interpretation of φ itself as an operator.Relativistic wave equations 9 i. there are solutions with negative energy.2) is j µ = i φ∗ ∂ µ φ or (ρ. And. This continuity equation can be written down o covariantly using the components (ρ.5). At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends.11) Therefore. relativistically the density is not a scalar quantity. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. 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. but rather the zero component of a four vector. This leads to the existence of negative densities. which is what Lorentz transformations do. ↔ ↔ (2.15) . however. just as the dependence on time was in ordinary quantum mechanics. As we will see later both problems are related and have to do with the existence of particles and antiparticles. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. (2π)3 2Ek (2. j) = ↔ i φ∗ ∂0 φ. The KG equation.8) (2. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2.13) ˜ with (in principle complex) coeﬃcients φ(k). there are no longer fundamental objections to mix up space and time. rather than as a wave function.14) ˜ ˜ Introducing φ(Ek . The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. (2. The appropriate current corresponding to the KG equation (see Excercise 2. k) + ei k·x φ(−Ek . −i φ∗ ∇ φ . 2. having the KG equation as a space-time symmetric (classical) equation. −k) . we want the most general solution. k) ≡ a(k) and φ(−Ek . ∂ρ = −∇ · j. (2.3). 2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. (2. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2.10) ∂t which follows directly from the Schr¨dinger equation. 2M 2M (2.9) They satisfy the continuity equation. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. j) of the four-current j.e.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x).

after restricting the energies to be positive).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) . It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.14 and 2. This shows how parity transforms k-modes into −k modes. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. 2.6 We will explicitly discuss the example of a discrete symmetry.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.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). x ∂µ ∂ µ + M 2 φ(˜) = 0. x) → (t. In terms of modes one has φC (x) = φ∗ (x) = ≡ d3 k e−i k·x b(k) + ei k·x a∗ (k) (2π)3 2Ek d3 k e−i k·x aC (k) + ei k·x bC∗ (k) . They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). (2π)3 2Ek (2.23) . Another symmetry is found by complex conjugating the KG equation. 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).15 one has elimated k 0 and in both equations k · x = Ek t − k · x. (2.e. φP (x) ≡ φ(˜) (P for parity). (2π)3 2Ek (2.16) The consequence of this is discussed in Exercise 2. Since a · b = a · ˜ it is easy to see that ˜ b. (2π)3 2Ek (2. (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). for which we consider space inversion.17) (2. −x) ≡ (˜µ ).Relativistic wave equations 10 In Eqs 2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). i. changing the sign of the spatial coordinates. (2. 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) .18) (2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. which implies (xµ ) = (t.

for the mode amplitudes aC (k) = b(k) and bC (k) = a(k). letting V → ∞. Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. d3 k F (Ek . 2.15.1 Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume. Note that a(k) and b(k) are independent of t. 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ). S V d3 x j 0 (x). Exercise 2.Relativistic wave equations 11 i. 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. |f ′ (xn )| . (2π)3 2Ek where Ek = Exercise 2. t) ≡ d3 x φ(x. k). Q = 0. ↔ ↔ Exercise 2. For the real ﬁeld one has aC (k) = a(k).3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . Exercises Exercise 2. ˙ and thus for any normalized solution the full ‘charge’.4 ˜ Write down the 3-dimensional Fourier transform φ(k. is conserved. t). t). t) = ei Ek t (i∂0 + Ek ) φ(k. ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. t) for the mode expansion in Eq. QV ≡ satisﬁes ˙ QV = − ds · j. k) = (2π)4 k2 + M 2 . that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B).e. t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k.

5 (a) As an introduction to parts (b) and (c). B] = AB − BA and the anticommutator as {A. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . 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 . i (a) Show that relativistic invariance. (c) Similarly.15.6 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). d3 x (∂ {0 φ)∗ (∂ i} φ). j 0 = ψ†ψ and j = ψ † αψ. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. 1 2 (b) Express now the full charge QV (t) (exercise 2. β} = 0. i. a{µ bν} ≡ aµ bν + aν bµ . From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function.Relativistic wave equations 12 Exercise 2.1) for a complex scalar ﬁeld current (exercise 2. Note that a{µ bν} indicates symmetrization. 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).7 To solve the problem with positive and negative energies and get a positive deﬁnite density. 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 . 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). . Exercise 2. {αi . αj } = 2 δ ij I.2) in terms of the a(k) and b(k) using the expansion in Eq. 2. also using the result in (a). in terms of the a(k) and b(k). ∂t where ψ is a wave function and the nature of α and β is left open for the moment.e. We will encounter these quantities later as energy and momentum. and β 2 = I. B} = AB + BA. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. t). One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). t) = (−i α · ∇ + m β) ψ(r. ∂ ψ(r. Exercise 2.

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

(3. Next. 0 (3. • [x. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. x]). which reﬂects itself in the noncommutation of the generators. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations.and y-axes 0 0 0 0 0 0 0 −i . [x. for rotations around z for instance. ] that satisﬁes • ∀ x. and thus (σ · n)2 = 1.5) Rotations in general do not commute.6) Summarizing. In the same way we can consider rotations around the x. [y. Locally this is a 3-dimensional Euclidean space and SU (2).4) R . z]] + [y. One way of viewing the parameter space. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. These generators form a threedimensional Lie algebra SO(3). Ly and Lz . y] ∈ A. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . x] = 0 (thus [x. z ) = lim N →∞ that are generated by i 0 . (3. y] = −[y. y ∈ A ⇒ [x. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n .2 π) (-n.1: the parameter spaces of SO(3) (left) and SU (2) (right).7) = a0 0 −1 i 0 1 0 0 1 • [x. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. y]] = 0 (Jacobi identity).Groups and their representations 14 (n. Lj ] = i ǫijk Lk . ˆ therefore. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. ˆ R(ϕ. is a 3-dimensional Lie-group. namely that there exists a bilinear product [. 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. 2 (3. with real a’s and 3 (ai )2 = 1. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). .8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). x]] + [z. [z. They satisfy the commutation relations [Li . π) (n.z N N or (Lk )ij = −i ǫijk .

ˆ ˆ A(ϕ. sometimes in more than one way (but this will be discussed later).12) One. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian.e. .e. n) ˆ −→ R(2π − ϕ. 3. The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. [g. thus. (3.11) Thus σx /2. however.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics.14) For a Lie group. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). it is suﬃcient that the generators commute with H. thus. n) i ∂ϕ = ϕ=0 (3. . n! 2! n=0 ∞ (3. (3. . for ﬁxed n.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . i. n) −→ R(ϕ. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. 2 2 2 (3. n) ≈ 1 + i ϕ J · n. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. Near the identity. . (3. They satisfy σi σj σk = i ǫijk . . 2 (3.10) σ ˆ · n.9) The parameter-space of SU (2).13) 0≤ϕ≤π The relation A(σ · a)A−1 = σ · RA a (valid for any vector a) can be used to establish the homomorphism (Check that it satisﬁes the requirements of a homomorphism. 2! . one has a Lie algebra isomorphism that is linear and preserves the bilinear product. In the full parameter space. n) π ≤ ϕ ≤ 2π. . immediately sees that the Lie algebras are identical. i. . also can be considered as a ﬁlled 3-sphere. i.15) [g. For instance in a function space translations. in particular that AB → RAB = RA RB ).e. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . SU (2) ≃ SO(3). H] = 0 for g ∈ G. the above mapping corresponds to the trivial mapping of the Lie algebras. Suppose we have a system described by a Hamiltonian H. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. H] = 0 for g ∈ G.

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

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

it is convenient to use a vector notation for the points in Minkowski space. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . 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 . We considered the generators and looked for representations in ﬁnite dimensional spaces.26) (3. e To derive some of the properties of the Lorentz group. G = 10 (3. Writing x as a column vector and the metric tensor in matrix form.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. e. 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.g.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. det Λ = −1 is called improper). Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.22) (Note that xT is a row vector). (ii) |Λ00 | ≥ 1.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.Groups and their representations 18 3. (3. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.25) (3. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .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 proper. From this property. σ/2 in a two-dimensional (spin 1/2) case. consisting of the Lorentz group and translations.27) (3.

3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. since the generators K do not form a closed algebra. 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).32) 0 − 0 −1 0 0 0 0 −1 .g. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. J = (J 1 . βγ = sinh η.Groups and their representations = i=1 (Λ0 )2 . K j ] = −i ǫijk J k . K 3 ). we have found the generators of rotations. K = (K 1 . Thus ˆ 0′ 0 0 0 V 0 V 1 0 0 0 0 1′ 1 V 0 cos ϕ sin ϕ 0 V 0 0 −i 0 2′ = 2 −→ J3 = V 0 − sin ϕ cos ϕ 0 V 0 i 0 0 . In L↑ . i proof: use ΛT GΛ = G and ΛGΛT = G. z ) ≈ ˆ ˆ I − i η K 3 . z ) = exp(i ϕ J 3 ). This is the origin of the Thomas precession. z ) ≈ I + i ϕ J 3 . K 2 . (3. inﬁnitesimally given by ΛB (η. Rotations around the z-axis are given by ΛR (ϕ. The commutator of two boosts in diﬀerent directions (e. J 3 ). [K i . [J i . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. it is easy to ﬁnd the following typical examples from each of the four classes. 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.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↑ . (3.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. and those of the boosts.28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞.30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3. one distinguishes rotations 0 + and boosts. z ) = exp(−i ηK 3 ). which satisfy the commutation relations (check!) [J i . K j ] = i ǫijk K k . Returning to the global group. Using these explicit transformations. 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. inﬁnitesimally given by ˆ ΛR (ϕ. J 2 . 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 . J j ] = i ǫijk J k .

Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .33) These four transformations form the Vierer group (group of Klein). a). 3 30 with η = ω and K = M (e. K = M ). ǫ). (3. i i0 i i0 1 ijk M jk 2ǫ (3.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .g. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). we continue using covariance arguments. inf (3.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. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 .36) (3. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. inﬁnitesimally approximated by (I + ω. In order to ﬁnd the commutation relations including the translation generators.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. This is a + normal subgroup and the factor group is the Vierer group. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ .37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . There are six generators. [M µν .34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. (3. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν .39) (e. The extension to the Poincar´ group is e ↑ straightforward. Summarizing. Also this group can be divided into four parts.g. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . Of the four parts only L↑ forms a group.38) Explicitly. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . P+ etc. We will just require Poincar´ invariance for the generators themselves. 3. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . We therefore ﬁnd (again) that there are six generators for the Lorentz group. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3.

The generators of the translations are the (momentum) operators P µ . as part of the set. leads to U (Λ. [P µ . = pµ |p. (3.38 could have been obtained in the same way. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. P j ] = i ǫijk P k .47) (3. P µ |p. the generators J± in the case of the rotation group. P j ] = [P i . one obtains 2 [P i . a)P µ U † (Λ. 3. [J i . [K i . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ .45) Note that the commutation relations among the generators M µν in Eq. Taking any one of the states in an irreducible representation. .46) [K i . = −i (g µρ P ν − g νρ P µ ) . These form a group called the little group associated with that four vector.41) 2 At this point. (3. Inﬁnitesimally. J j ] = i ǫijk J k . (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . ǫ) = 1 − ωαβ M αβ + i ǫα P α . H] = [J i .44) (3. (3. [J i . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. the generators M αβ no longer exclusively act in Minkowski space. For instance. Explicitly. a)P µ U (Λ. .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . They just state that M µν transforms as a tensor with two Lorentz indices. H] = i P i . .42) (to decide on where the inverse is. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . known from non-relativistic quantum mechanics are obtained.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. P ρ ] = 0. . P j ] = i δ ij H/c2 . H] = 0. We have here reinstated c. check that U1 U2 corresponds with Λ1 Λ2 ). other states in that representation are obtained by the action of operators outside the maximal commuting set. K j ] = i ǫijk K k . For instance. The eigenvalues of these will be the four-momentum pµ of the state. (3. 3.Groups and their representations 21 inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. . that commute e among themselves. ν (3. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . [J i . . P ) in terms of the Hamiltonian and the three-momentum operators. . These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. 2 2 (3. a) = Λµν P ν or U † (Λ.48) . K j ] = −i ǫijk J k /c2 . [K i . the generators J 2 and J 3 in the case of the rotation group. a) = Λ−1 µ P ν . P ν ] [M µν . writing the generator P = (H/c.

. From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. p0 > 0. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p.56) . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). .55) commute with all generators and therefore are invariants under Poincar´ transformations. 2 (3.52) (3. . 0. the third case represents the vacuum. . i ǫµνρσ W ρ P σ . Pν ] = [Wµ . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν .53) (3. = = i exp − ǫµνρσ M µν pρ sσ |p. Wα ] = [Wµ . while the others have no physical signiﬁcance. Massive particles: p2 = M 2 > 0. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . . 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. .51) The following properties follow from the fact that W µ is a four-vector by construction. (3. which can thus be chosen spacelike. Wν ] = −i (gµα Wν − gνα Wµ ) . .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . (3. M ǫijk .54) This will enable us to pick a suitable commuting ’spin’ operator. 2 exp (−isµ W µ ) |p. 0.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. . .49) (3. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3.

Together with the four momentum states thus can be labeled as |M.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. M2 (3. M J) with components W i = (M/2)ǫijk M jk . p. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 1. W j (p)] = i M ǫijk W k (p). 0).57 The best insight in the meaning of the operators W µ is to sit in the rest frame of the particle and put P µ = (M. |p3 |). We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. 0). n1 (p). In that case the vectors ni (p) are just the space directions and W = (0. W i (p) = −W · ni (p)). 1. . M2 (3.. p0 > 0.e. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. show that W /M can be interpreted as the intrinsic spin. 3. 0. (1.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 .61) one sees that S i (p) i =− W2 .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).64) such that in an arbitrary frame where the four vectors p. i (3.e. 1 2 (3. The commutation relations [W i (p). 0. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. 0. 0. 2 . (3. . 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. s. 0). 0.59) (3. ms . 1. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). (0. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p).57) (i. Having made a covariant decomposition. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). .58) i. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. (0. (3.65) .

a)|0 = |0 . |p. The vacuum: pµ = 0. i W 2 (p). where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p).66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). 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. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). and gives [W 1 (p). But if M = 0 one has a diﬀerent algebra (Eq. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. |p| (3. and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0). Note that angular momentum does not contribute to helicity as L · p = 0. (3. W 2 (p)] = = = i M 2 λ(p). Exercises Exercise 3. Thus we have P 2 .70) as a commuting set with zero eigenvalues for P 2 and W 2 . however. 3. is characterized by its momentum and the helicity. The algebra of W 1 . which can take any integer or half-integer value. thus. W 2 and λ is derived from the general commutation relations for [Wµ . σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a).71) which is called the helicity.13. (3. This state is in physical applications nondegenerate and is denoted by |0 .68) (3.67) (3. U (Λ. 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). implying continuous values for the spin (actually for W i = M S i ). Wν ]. W 1 (p)] [λ(p).1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. P µ . 3. A massless particle.73) (3. We don’t know of any physical relevance.Groups and their representations The above does actually include the massive case.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. W µ (p) = λ(p) P µ . Thus 2 λ(p) = J ·p . −i W 1 (p). W 2 (p)] [λ(p). 1 2 24 (3. W 2 (3. λ .67 has a vanishing right hand side). 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 .

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

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

C) is the group of complex 2 × 2 matrices with determinant one. It is simply connected and forms the covering group of L↑ .2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). This tells us how to ﬁnd the representations of the group. K = −i . C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . C) can be written as a product of a unitary + matrix U and a hermitean matrix H. They will be labeled by two angular momenta corresponding to the A and B groups.1 The Lorentz group and SL(2. K = +i . respectively. (j. The group SL(2. (4. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. i ǫijk B k .Chapter 4 The Dirac equation 4. 0. becoming the deﬁning representations of SL(2. They precisely correspond to the two types of representations that we have seen before.3) (4.7) From the considerations above.6) (4.9) (4. U (ϕ) H(η) (4. j ′ ). 0) Type II : (0. Aj ] = [B i .8) ˆ ˆ ˆ with η = η n and ϕ = ϕn.10) . The matrices in SL(2. 2 1 (J − i K). 2 2 σ σ J = . 2 (4. Special cases are the following representations: Type I : (j.5) = 1 (J + i K).4) (4. With this choice of parameter-spaces the plus and minus signs are actually relevant. B j ] = i ǫijk Ak . where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. 2 2 (4. K = i J (A = 0). (4. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . C). 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 . j) K = −i J (B = 0).1) (4. similarly as the homeomorphism between SO(3) and SU (2). B j ] = [Ai .

cannot be connected by a Lorentz transformation belonging to L↑ . Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . In fact. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. ˜ = σ µ Λ B aµ . With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. C). One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . such that M = SM S −1 . (4.e.12) ξ → U ξ. thus. while ±ǫη ∗ transforms as a type-I + spinor. ǫ−1 = ǫ† = ǫT = −ǫ). They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that y + ↑ ∗ ǫ ∈ SL(2.16) = σ µ Λ R aµ . This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. (4.11) Considering ξ and η as spin states in the rest-frame. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). H(η) = exp(−η · σ/2).19) . C).15) (4. σ µ ≡ (1. C) or L↑ . η → U η.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices.1). i. Eqs 4. σ). namely H(η) = exp(η · σ/2). Λ(ǫ) ∈ L+ . This shows that M ∗ ≃ M . The following relations for + rotations and boosts are useful. ξ −→ Hξ. there exists the matrix ǫ = iσ 2 .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. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. (4. if there does not exist a unitary matrix S. The Lorentz transformations Λ(M ) and Λ(M ). η → Hη. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. C) and. the ﬁrst equation shows that the conjugate representation with spinors χ∗ transforming according to −σ∗ /2 is equivalent with the ordinary two-dimensional representation (±ǫχ∗ transforms according to σ/2). This is not true for the two-component spinors in SL(2.18) As already discussed for SU (2). H ≡ (H † )−1 = H −1 ).13) 2M (E + M ) (exercise 4.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. −σ). One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . ˜ (4. (ǫ∗ = ǫ. (4.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. however. thus. U σ µ aµ U −1 = σ µ ΛR aµ . (4. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. one can use a boost to the frame with momentum ˆ ˆ p.

they cannot be connected by a unitary 2 2 transformation. 4. i. 4. (axial vector).13. The representing member of the class P− is the parity or space inversion operator Is . A parity transformation changes aµ into aµ .g. From the deﬁnition of the representations (0. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. The behavior is the same for classical quantities. ) −→ ( . (4. but to L↑ . the boost in Eq. (pseudoscalar). −→ −r (vector). but by a transformation belonging ↑ ↑ to the class P− . −a). m) = χm . 0) of SL(2. C) are suitable for representing spin 1/2 particles. For the spin 1/2 representations of the Poincar´ group including parity we. = (4.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. −→ −p (vector). consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . but rather the operators W i (p) should be used. −→ M (Λ) (4. Although these operators cannot be used to label the representations. m) = η(0. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). 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 2 (4.The Dirac equation 29 where The matrix I2 does not belong to L↑ . 0) are inequivalent. 2 ) and ( 2 . e.17. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle.23) u(p. 1 ) and ( 1 . i.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0.20) 4. (vector).e. 0). This provides the easiest way of seeing ˜ what is happening. therefore. It has the same eﬀect as lowering the indices. as we have seen in section 4. we have seen that in the rest frame W i (p)/M → J i . The i angular momentum operators J are represented by σ i /2. the well-known two-component spinor for a spin 1/2 particle. For a particle at rest only angular momentum is important and we can choose ξ(0. 1 ) and ( 1 . etc. where a = (a0 . Thus M ∗ and M are not equivalent within SL(2. generators. the aij elements of a tensor will not change sign.e. Within the Lorentz group. they can be connected.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. m) = χm uL H(p) 0 . must combine the repe resentations. C). ǫµνρσ will change sign. a) and ˜ a = (a0 . We also have seen that the representations (0. 0 χm uR H(p) .1. (scalar). that have the correct commutation relations.

= (4.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .27 is known as the Weyl representation. {γ µ . To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). From this one obtains the Cliﬀord algebra for the Dirac matrices.28) which is a covariant (linear) ﬁrst order diﬀerential equation.33) deﬁne for a four vector a the contraction a = aµ γµ / . (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. 2 (4.7. We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0.29) which is an explicit realization of the (momentum space) Dirac equation. (4. The general requirements for the γ matrices are thus easily obtained. . this can be rewritten 1 µ ν {γ . (4.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). We will discuss another explicit realization of this algebra in the next section. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. γ } ∂µ ∂ν + M 2 ψ(x) = 0. = H −1 (p). The explicit realization appearing in Eq. 4. The ﬁrst indeed is solved.26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0. Since ∂µ ∂ν is symmetric. It is of a form that we also played with in Exercise 2.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. (4. Applying (iγ µ ∂µ + M ) from the left gives (4. γ ν } = 2 g µν . 0 H 2 (p) uR uR . (4. p (4. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). which in general is a linear equation in pµ . This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. (iγ µ ∂µ − M ) ψ(x) = 0.24) (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 .25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. (4.32) suppressing on the RHS the identity matrix in Dirac space.

1: The Dirac sea of negative energy states and antiparticles. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).g. This can most easily be seen in the particle rest frame.34) for the adjoint spinor ψ ≡ ψ † γ0 .35) is indeed positive and j 0 can serve as a probability density. opposite charge). Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0.27).39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. (4.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. (4. The second problem encountered before. 4. . but with properties such that it can be annihilated by the particle (e. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. 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. remains. From the vacuum a particle can be removed. The density j 0 = ψγ 0 ψ = ψ † ψ (4. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. γ ν } = 2 g µν . (4.37) Using the explicit form of γ in the Weyl representation (see Eq. Relying on the Pauli exclusion principle for spin 1/2 particles. The Dirac sea forms the vacuum. . This hole forms an antiparticle with the same mass. the one of the negative energy states. This problem was solved by Dirac through the introduction of a negative energy sea.36) (4. 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 . γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4.3 General representations of γ matrices and Dirac spinors {γ µ . E = +M (twice) and E = −M (twice). one sees that there are two positive and two negative eigenvalues.

the rotation and boost generators in Weyl representation in Eqs 4. γ 2 ]. 0 (4. −→ Sγµ S −1 .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.3) .R. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k .R. (γµ )S. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. We note that ψ → Lψ and ψ → ψ L−1 .41) (4. For instance in the Weyl representation (but valid generally).47) (4.45) (γ5 )W. while L−1 γ µ L = Λµ γ ν . γ µ } = 0 and explicitly one has 0 1 1 .9 and 4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . 2 (4. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . = S(γµ )W. that project out chiral right/left states. −→ Sψ. γ σ ]. =ρ ⊗1= 0 −1 3 (4. (γ5 )S.43) S=√ . 4! (4. Another matrix which is often used is γ5 deﬁned as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisﬁes {γ5 . (4.R. is 1 1 1 (4.46) (4. For example. γ 0 ]. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 . Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. S −1 .R.48) are projection operators.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 .40) (4. which in the case of the Weyl representation are just the upper and lower components.

= = 0 iγ γ = −iρ ⊗ σ = −iσ 2 −iσ 2 iγ 2 γ 0 = −iρ3 ⊗ σ 2 = 0 2 0 1 2 Thus we ﬁnd back the Dirac equation.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. ∂ (4. usually to be taken unity. reverse the charge of the particle. x) → x = (t.54) e. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. Explicitly. called charge conjugation. −iσ 2 0 −ǫ = . x = (t.55) (4.59) −→ ψ c = C ψ = η ǫ ξ∗ . 0 −ǫ 0 0 −ǫ 0 . = 0 ǫ iσ 2 (4. ∂ with the solution ψ c (x) = ηc Cψ (x). (i/ − M ) ψ(x) = 0.R. (C)S. (or W. T (4.53) (4. γ5 ] = 0. T (4. such that T Cγµ C −1 = −γµ .50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. ψR . but it does. is a left-handed spinor.g. (C)W. C is real and one has C −1 = C † = C T = −C c and [C. ψ= (4. for instance.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation.52) η ξ one can apply space inversion. As with parity we look for a transformation. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ .) and all representations connected via a real (up to an overall phase) matrix S. This symmetry does not change the spin 1/2 nature.58) where ηc is an arbitrary (unobservable) phase.R. ψ= (4. Some properties of C are C −1 = C † and C T = −C. Starting with (i/ − M ) ψ(x) = 0 we note ∂ that by hermitean conjugating and transposing the Dirac equation one obtains iγ µT ∂µ + M ψ (x) = 0.56) (i/ − M ) ψ c (x) = 0.6).R. One has ψ c = −ψ T C −1 . parity. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.57) (4. The latter implies that the conjugate of a right-handed spinor.R. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η .g. In any representation a matrix C exist. that brings ψ → ψ c . e. which is again a solution of the Dirac equation. ∂ (4. In S.

p (4. s)¯(p.64) (4. ∂ −iσ · ∇ −i ∂t − M (4. s) u = . ¯ ′ ′ Explicit solutions in the standard representation are χs . In that frame is a spacelike unit vector sµ = (0. ⇔ (/ − M ) u(p) = 0.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. where v satisﬁes σ·p −(Ep + M ) v(p) = 0.69) = − s In order to project out spin states. The ¯ ¯ solutions are normalized to u(p. s) = u(p. s)v(p. with E = Ep = + p2 + M 2 .61) σ·p −(Ep + M ) There are also two negative energy solutions. s) and C v T (p. be obtained by a Lorentz transformation.g. 2M 2M −/ + M p v(p. the spin polarization vector in the rest frame is the starting ˆ point. s′ ) = 2M δss′ . It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4..67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p.68) (4. s) . 2M 2M (4. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s)v(p. s) = u(p. v (p. the best representation to describe particles at rest is the standard representation. the spinors satisfy C uT (p. ¯ (4. s)u(p. (4.63) u(p.70) = Ps = ˆ 0 1∓σ·s 2 2 . s)u(p. s)v(p. s)u(p. s)¯(p. ψ(x) = u±s (p) e−i p·x . s).The Dirac equation 34 4.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4.65) (4. In an arbitrary frame one has s · p = 0 and s(p) can e. where u satisﬁes −σ · p Ep − M u(p) = 0. 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. ¯ ¯ † ′ u (p. s′ ) = 2Ep δss′ .60) ψ(x) = 0. v(p. s) + v(k. s) ei k·x d∗ (k. s) e−i k·x b(k. s ) = 0.62) where χs are two independent (s = ±) two-component spinors. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s). s ) = v † (p. (2π)3 2Ek (4. ψ(x) = v ±s (p) ei p·x . Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). ⇔ (/ + M ) v(p) = 0. s′ ) = −2M δss′ . s) v = . s) = Ep + M σ ·p Ep +M χs . s ) = v (p.1). in which γ 0 is diagonal (see discussion of negative energy states in section 4. s) = v(p.

helicity states.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . 1 u(p. λ). In principle massless fermions can be described by twocomponent spinors. 4. . Therefore. λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. v(p.74) . u(p. +1/2) = √ u(p.The Dirac equation 35 (the last equality in the restframe and in standard representation) projects out spin ±1/2 states (check ˆ ˆ this in the restframe for s = z ). For instance with the z-axis as spin quantization axis. The chirality projection operators in Eq. λ = − 1 ) = √ √ v(p. while the negative helicity (λ = −1/2) is in essence lefthanded. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . λ = − 1 ) = √ E + M − E − M . λ = −1/2) = √ − E+M √ 0 0 E+M (4. Note that for solutions of the massless Dirac equation /ψ = 0. v(p.73) Also for helicity states C uT (p. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. Explicit examples of spinors are useful to illustrate spin eigenstates. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). (4. λ = +1/2) = v(p. p ψR/L ≡ PR/L ψ are independent solutions.48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). one has in standard representation: 0 E+M E+M 1 1 0 . −1/2) = √ (4. This means that in the solution space for massless fermions the chirality states. λ) = u(p. 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 √ . +1/2) = √ 0 E +M E+M . −p3 −p1 − i p2 1 v(p. 2 2 0 √ 2 √ 2 − E+M − E−M 0 . λ = + 1 ) = √ + E +M − E −M . λ = −1/2) = u(p. √ 0 u(p. (4.71) p − i p2 . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. −1/2) = √ E + M −(E + M ) 0 . λ) = v(p. λ) and C uT (p. etc. chirality. u(p.

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

g. of which only the traces of four and two gamma-matrices are nonzero. −x) is also a solution of the Dirac equation. γ σ ] = σ ρσ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . s) and v(p.13.g. s)¯(p. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . 1 i −i S ρσ = − [γ ρ . γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. 4.77) Exercises Excercise 4. Exercise 4. Show that the equality holds. H(p) = exp η·σ 2 = M +E+σ·p . where x = (t. a 4 a · b. ˜ Exercise 4. Do this e.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4.5 Show starting from the relation {γµ . x ˜ . x → x. Excercise 4. 4 2 Exercise 4. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜).1 Prove Eq. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. (4. by letting them act on the four basis spinors u(p. s) u p /+M = 2M 2M both sides are projection operators. s). i S ρσ ] = g ρµ γ σ − g σµ γ ρ .6 Apply space-inversion.4 Show that in P+ ≡ s=±1/2 u(p.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. 0 if n is odd. e.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν .

57 MeV.The Dirac equation 38 Exercise 4. namely i / ψ = eiαγ5 ψ. The pion has spin 0. r).8 × 10−6 . Calculate for now just |M |2 . the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. Do you understand why? The calculation of the life-time is completed in chapter 11. mµ = 105. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. there is no current ﬂowing through the surface of the cavity. Excercise 4. Note: To prove this. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139.5 and M → ∞. me = 0.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e .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 ). The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. Calculate the value of k for M R = 0.66 MeV. We will not pursue this chiral freedom at this point and take α = 0. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. we want to conﬁne this solution to a spherical cavity with radius R. For such a cavity nµ = (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 ). i. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. the decay would even be forbidden. . M R = 1. 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. n It turns out that there is actually a whole family of conditions that provide conﬁnement. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). you need to determine the corresponding conﬁnement condition for ψ. (b) Next.511 MeV). ν p Note: how such parametrizations work will be explained in chapter 8. is a stationary solution of the (free) Dirac equation. 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.

ǫλ ).1 Fields for spin 1 In section 3. where k = k0 = (M.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. M2 (5.2 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. In the restframe. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .6) where Aµ is the four-vector potential (5. with ∂µ V µ = 0 (Lorentz condition).5) −E B2 . 39 (5. These are solutions of the Klein-Gordon equation with in addition a condition. These can describe spin 1 states in the rest-frame. λ) . there are then indeed three independent possibilities. (2 + M 2 )Vµ = 0.Chapter 5 Vector ﬁelds and Maxwell equations 5.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k.2) (5. (5. namely ǫµ (k0 ) = (0. ±(1) are obtained by boosting the restframe vectors. just as the two-component Dirac spinors did for spin 1/2.1) The Lorentz condition implies that k µ ǫµ (k) = 0.7) . 0).4) λ=0.± 5.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. In an arbitrary frame ǫµ (k) with λ = 0. λ) + ǫ(λ)∗ (k) eik·x c∗ (k. µ (2π)3 2E µ (5. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. The vectorﬁeld is therefore (λ) suited to describe spin 1. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. A).

40 (5. transverse or Coulomb gauge). Taking the divergence of this equiation.11) This equation is not aﬀected by a gauge transformation. (5. j) the Maxwell equations read ∂µ F µν = j ν .14) of which the solutions give the Li´nard-Wiechert potentials. Furthermore. −i. if ∂µ Aµ = 0 for large times |t|. ∓1.17) (radiation. Therefore if k µ = (|k|.8) (5.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. i. 2Aµ = 0. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . in which case one has 2Aµ = jµ . One possible choice is the gauge A0 = 0 (5. ˜ x Note that via parity transformation one obtains another solution. (5.10) the electric and magnetic ﬁelds are unchanged. (5.e. 5. The equation in vacuum. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. 0). 0. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. (5.13) Hence. moree over. It states the absence of magnetic charges and Faraday’s law.15 the d’Alembertian is replaced by 2 + M 2 .5.15) (5.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. . only two independent vectors remain ǫ(±) µ (k) = (0. As remarked above. if in Eq. AP (x) = Aµ (˜). |k|). it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. (5.16) (5. For electrodynamics one has furthermore the freedom of gauge transformations. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. 0. Although the Lorenz condition is a constraint. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. in accordance with what we found in section 3. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . for M = 0. ∂µ Aµ = 0. it will vanish at all times.12) (5. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x).

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

it can be obtained from the situation A = 0 by a gauge transformation. The AharonovBohm experiment shows a realization of this possibility.21) (ϕ being the azimuthal angle). is multivalued as ϕ = 0 and ϕ = 2π are the same point in space.2: The topology of the space in the Aharonov-Bohm experiment. This situation that A can be written as ∇χ is always true when ∇ × A = 0. A = ∇χ with χ= BR2 ϕ 2 (5. 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. i. Another situation in which the topology of space can be used is in the case of superconductors. giving no observable phase2 . Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. 5. minimizing the space occupied by B ﬁelds.e. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. 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. Now back to the Aharonov-Bohm experiment. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . If this would be be happening in empty space one would be in trouble. ϕ=2π The function χ in Eq.21. 1 The occurrence of nontrivial possibilities. implying that it must be single-valued1 . however.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. In such a space loops with diﬀerent winding number (i. Consider a superconductor. Therefore space outside the solenoid doesn’t care that χ is multi-valued.e. forming a simple (connected) space. the number of times they go around the defect) cannot be continuously deformed into one another. nonobservable phases ϕ = 2π n. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). The only way for ﬂux tubes to be formed and move around without global consequences is when each ﬂux tube contains a ﬂux Φ such that ∆δ = (e/ )Φ = n · 2π. Therefore the gauge transformation must be uniquely deﬁned in the space one is working in. organizing itself in tiny ﬂux tubes (Abrikosov strings).

∂t ∇ · B = 0.Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5. .1 Show that Eqs.8 and 5. ∂t ∂B = 0. 5. ∇×B =j+ ∇×E+ ∂E .9 explicitly give the equations ∇ · E = ρ.

x) ˙ (6. ∂x (6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation).2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. As an example. recall classical mechanics with the action t2 S= t1 dt L(x. x) = K − V = ˙ 1 mx2 − V (x). ˙ 2 (6. These equations can be obtained from a lagrangian using the action principle. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. (6. d dt ∂L ∂x ˙ = ∂L . x (t) = x(t) + δx(t). thus t′ = t + δτ. and the total change x′ (t′ ) = x(t) + ∆x(t).5) with ∆x(t) = δx(t) + x(t) δτ .1) and as an example the lagrangian L(x. ∂x ˙ (6.8) 44 .3) (6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. ˙ p= ∂L .6) The ﬁrst term leads to the Euler-Lagrange equations.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.Chapter 6 Classical lagrangian ﬁeld theory 6.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 . t1 (6.

with ∆φ(x) = δφ(x) + (∂µ φ)δx . dσµ . . x′ ) − d4 x L (φ.12) Here R indicates a space-time volume bounded by (R3 .13) (6.14) (6.2.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . their derivatives and possibly on the position. = ∂R (R3 . − d3 x . x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). The resulting variation of the action is δS = R µ ′ = xµ + δxµ . . i. t1 ) and R3 . . 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 . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. ˙ (6. ∂µ φ(x)) = R d4 x L (φ(x). think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). . In general any continuous transformation. λ=0 (6. but which in 3 dimensions includes conserved quantities related to rotations and boosts. ∂µ φ(x). ∂µ φ(x)). also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered).17) Furthermore the variations δφ and δ∂µ φ contribute to δS.e. .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). 6. . = φ(x) + δφ(x) (6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). . (R3 .11) λ=0 of which p and H are the simplest examples related to space and time translations. ˙ x The second term in Eq. H is a conserved quantity. x). Consider a lagrangian density L which depends on these functions.Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. Variations in the action can come from the coordinates or the ﬁelds.15) d4 x′ L (φ′ . We will come back to it in a bit more formal way in section 9.18) 1 Taking a functional derivative. ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . t2 ). + ∂L ∆x − H δτ ∂x ˙ t2 . ∂µ φ. t1 (6. (6. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . L (φ(x).6 can be rewritten as δS = . The hamiltonian H is deﬁned by H(p.t2 ) d3 x .9) which is done because the ﬁrst term (multiplying ∆x. which in classical mechanics vanishes at the boundary) does not play a role. ∂µ φ′ . ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. x) ≡ p x − L . δ(∂µ φ) (6. R (6. . . indicated with δF [φ]/δφ should pose no problems. .t1 ) (6.

(6. leading to (2 + M 2 )φ(x) = 0.26) 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.20) (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). the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. One easily obtains (2 + M 2 )φ(x) = 0. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. Using the variations in ψ (in the symmetric form).21) which diﬀer only by surface terms. The integrand of the ﬁrst term must vanish. ∂ (6. (6. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). 2 ∗ (6.29) .25) (6. δL δL ∂µ = . the second term is irrelevant.24) (6.23) (6.19) δ(∂µ φ) δφ 6. (2 + M )φ (x) = 0.2 Lagrangians for spin 0.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. 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. 2 ψ = 0. leading to the Euler-Lagrange equations. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. (6.27) (6. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities.

Peshkin and Schroeder or Exercise 12. and thus (β ∗ α)∗ = α∗ β = −βα∗ . which comes because we in essence use ’real’ spinors. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).g.32) There exists another possibility to write down a mass term with only one kind of ﬁelds. 2 αβ = −βα. 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. We note that of which the left part coincides . (6. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . (6.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.38) (6.31) showing e.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.31 is given by LM (Dirac) = −M ξ † η + η † ξ . that the lagrangian separates into two independent parts for M = 0. 2 (6. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).36) (6. 1 M η † ǫη ∗ − M ∗ η T ǫη .30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . The reasons for Grassmann variables will become clear in the next chapter. (6.34) (6. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . ψL ≡ (ψL )c = C ψL = 0 (6. The Majorana case is in fact more general.g. or equivalently separate the ﬁeld into right.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . ∂ ∂ 2 2 (6. Using the twospinors ξ and η.7). the mass term in the Dirac lagrangian 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 . but note the factor 1/2 as compared to the Dirac lagrangian. . namely2 LM (Majorana) = + (6.and lefthanded ones. 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.

Q(t1 ) = Q(t2 ). which in a consistent picture precisely generate the symmetries.44) δL ∆φ + Θµν (x)δxν . (6. + ∂R dσµ dσµ ∂R ≡ where d x .. discussed in the next chapter.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν .45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . . Q= d3 x J 0 (x.40) 6.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x).. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. we consider the second term in Eq.18. . these conserved quantities become operators. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. (6. the presence of a symmetry that leaves the lagrangian invariant. which is the space-integral over the zero-component of a conserved current. (6. In particular when the lagrangian is invariant under symmetries. t). δ(∂µ φ) For continuous transformations. These do not aﬀect the dynamics and will give rise to conserved currents.3 Symmetries and conserved (Noether) currents Next. one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0.46) For quantum ﬁelds (φ operator!) things become more subtle as it it is not possible to specify ˙ for instance φ and φ on the surface σ1 .47) . ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) .Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . In this case. As an example for the classical ﬁeld. ∂µ J µ (x) = 0. requires the presence of a conserved quantity. 6. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. one can consider the variations at the initial or ﬁnal surface. δ(∂µ φ) (6.41) (6. (6. (6. Returning to δS the surface term is rewritten to δS = R d4 x . t). δ(∂µ φ) (6.σ1 with σ = (R3 .

H = d3 x Θ00 (x).51) (6.3) j µ = i e φ∗ ∂µ φ. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . (6. 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. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).53) These are the energy and momentum. (6. 1 (6. = 0. including e the space-time translations and the Lorentz transformations. (6.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations.52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0. The integral over the zero-component. e.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld. ˙ Q = d3 x j 0 (x. From Noether’s theorem one sees that the current Θµν ǫν is conserved. ↔ (6. .56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . (6.49) These currents are conserved as discussed already in chapter 2.54) (6. In the next section we will see the quantity show up as the charge operator. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds.50) (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). Summarizing. 2 (6.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters. we obtain (since δxµ = 0). Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). 6.g. t) is the conserved charge (Q = 0).

3). Any lagrangian containing derivatives. (6.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0.67) i. ∂µ φ(x) + i e ∂µ Λ(x) e (6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .58) (6. ∂µ T µν (6.3 we have seen global transformations or gauge transformations of the ﬁrst kind.64) (6.59) (6.68) (6. however. −i eΛ(x) ∗ i eΛ(x) (6.63) µν = ∂µ Θ .69) i eΛ(x) φ (x). γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. (6. (6. e. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). In general this tensor is not symmetric. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. where H ρµν is the quantity appearing in M ρµν . • Vector ﬁeld Aµ : −i Sρσ = aρσ .g.e. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). independent of x. φ(x). 2 (6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.65) M µρσ = T µρ xσ − T µσ xρ . 6.60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . one has a tensor that satisﬁes (exercise 6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. Deﬁning T µν = Θµν − ∂ρ Gρµν . (6. the angle of the transformation depends on the space-time point x. in which the U (1) transformation involves an angle e Λ.6) T µν = T νµ .70) .5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).61) A ﬁnal note concernes the symmetry properties of Θµν . is not invariant under local gauge transformations. (6.66) In section 6. 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 .

Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x).77) involving the interaction of the charge (ρ = j 0 ) and three-current density (j) with the electric potential (φ = A0 ) and the vector potential (A). (6.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. ∂ 2 4 (6. 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. 4 (6.78) . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. (6. 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.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. D ∂ A (6. ∂µ φ) =⇒ L (φ.72) (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . 1 L (φ.73) (6.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. As an example consider the Dirac lagrangian. (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 . −e j µ Aµ = −e ρφ + e j · A.74) the behavior which we have encountered before as a gauge freedom for massless vector ﬁelds with the (free) lagrangian density L = −(1/4)Fµν F µν . Dµ φ) − Fµν F µν . 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν .

e. 4 (b) What is the form of the interaction term involving a current jµ and the ﬁeld Aµ that will give the inhomogeneous Maxwell equations.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . give the equation which is satisﬁed by ψ c = Cψ . ↔ Exercise 6.79) Exercises Exercise 6. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). and deduce what is the charge of the antiparticle as compared to a particle. i. φ → eiΛ φ.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. −eψγ ν ψ. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. 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 6.1 (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 . . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. ∂µ F µν = j ν . giving the Maxwell equation coupling to the electromagnetic current. ∂µ F µν = j ν . (6. Exercise 6.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν .3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds.

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

(7. labeled by the position. ˙ L(q. The bilinear product for the conserved quantities. (7. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. ˙ 2 (7. 1 qx (t) = 3 d3 x φ(x.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. (7. starting from the lagrangian L(q. 6. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. namely [q. p] = i. q(q.3) For variations δO(p. q) also considered in the previous chapter. H]P dt and dO = [O. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. q) = p q − L(q. t). p]P = 1. however. Further. B]P = dB dA dA dB − . Q]P . q) = ˙ 1 mq 2 − V (q). p]P .11) found through Noether’s theorem. dq dp dq dp (7. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. there are conserved ˙ ˙ quantities Q (Eq.Chapter 7 Quantization of ﬁelds 7.2) H(p. [q. dλ Examples are dO = [O. An immediate generalization for ﬁelds can be obtained by considering them as coordinates.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). in our example p = mq.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t). is the Poisson bracket [A.5) ∆ x ∆3 x 54 .

for symmetry transformations. Φ(f ) = d4 x φ(x) f (x).14) In fact. (7.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. t)] = [Π(x. Sometimes it is useful to keep in mind that. Π(y. ∂ qx ˙ ∂ qx ˙ (7. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). ∂µ φ(x)) = ˙ d3 x L (φ. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). discussed in the section 6. qx+∆x ) . a) = Rj (Λ−1 ) φj (Λ−1 x − a).Quantization of ﬁelds 55 etc.11) (7.12) (7. The essential conditions to consistently quantize a theory then are the following. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x.6) ∆3 x L x (qx . (7.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.16) . t)] = i δ 3 (x − y).8) (7. (7. φ(y. ˙ δ φ(x) (7. albeit with ’sharp’ functions.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . 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. the discretization procedure above is an explicit example.15) with furthermore the relations [φ(x. qx . the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . φ. formally. for the scalar ﬁeld theory. ∇φ) (7. 2 2 2 (7. t).7) d3 x L (φ(x). ∆φ. a) φi (x)U (Λ. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . • Poincar´ invariance e The full variation of the ﬁelds. t).2. Π(y. i U † (Λ. t). The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). t)] = 0. As an example.

17) (Sµν )i φj (x). The measurements cannot inﬂuence each other or [Φ(f ). Microscopic causality implies local commutativity.24) It is straightforward to ﬁnd the commutation relations between N and a and a† .25) .22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. (7. j 2 one has [φi (x). In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . [φ(x). 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. Q] = [P. φ(y)] = 0 if (x − y)2 < 0. sometimes referred to as second quantization.18) with S µν given in 6. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). and [N. a] = [a† . [N.Quantization of ﬁelds 56 or if U (Λ.4. [Q. ω a† a + 2 2 ω i − [Q. a† ] = a† . (7. P ] = 0 (7. a† ] = 0.2 Creation and annihilation operators Before discussing (real and complex) scalar ﬁelds and Dirac ﬁelds we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. 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. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν .20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. Φ(g)] = 0. j (7. P ] = i. Mµν ] = i i∂µ φi (x). Pµ ] = [φ (x).21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . 2 i ωµν (S µν )i .19) 7. [a. a† ] = 1. 2 2 (7.23) The hamiltonian in this case can be expressed in the number operator N = a† a. P ] 2 (7. (7.

4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. a(k ′ )] = [a† (k). φ(x )]x0 =x′0 = [Π(x). (2π)3 2Ek (7. 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 .28) (7. is equivalent with [a(k).30) [φ(x). i. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). which represents the ’number of particles’ with momentum k.like’ commutation relations between a(k) and a† (k ′ ). Note that we will often write a(k) or a† (k). (7.26) a(k) e−i k·x + a† (k) ei k·x . N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . † √ a and a act as raising and lowering operators.33) .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. and (7. [φ(x).e. and we see that a state |0 must exist for which N |0 = a|0 = 0. In this way one has found for the harmonic oscillator the spectrum of eigenstates |n (with n a non-negative integer) with En = (n + 1/2)ω. (n − 1) a|n † i. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.g. 7. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.32) (7.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . the real scalar ﬁeld φ(x) becomes d3 k (7. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .e. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. Π(x )]x0 =x′0 = 0.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . e. a† (k ′ )] = 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. . (2π)3 2Ek (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. (2π)3 (7. . .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.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). |0 . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .42) (7. deﬁning particle states |k = |k (with positive energy.40) This procedure is known as normal ordering. (2π)3 2Ek (7. k 0 = Ek = k2 + M 2 ). d3 k a(k) e−i k·x .38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).44) p|φ− (x)|0 = ei p·x . (2π)3 2Ek (7. (2π)3 2Ek (7. is solved by subtracting it as an (inﬁnite) constant.e.45) (7.46) (7. |k = a† (k)|0 . 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). which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). assuring that the vacuum (by deﬁnition) has eigenvalue 0!. . writing all annihilation operators to the right of the creation operators. d3 k |k k|. 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).34) where V = (2π)3 δ 3 (0) is the space-volume.39) The problem with the vacuum or zero-point energy. i. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x).37) (7. The rest of the states are then obtained from the groundstate via the creation operator. (2π)3 2Ek d3 k a† (k) ei k·x . which now contains an inﬁnite number of oscillators. n1 ! n2 ! (7. a(k)|0 = 0. and multiparticle states |(k1 )n1 (k2 )n2 .35) where the vacuum contribution disappears because of rotational symmetry. This term will be adressed below.41) (7. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . .

(2π)3 2Ek ≡ i∆− (x − y). ∆(x) = − ǫ(x0 ) δ(x2 ). that they satisfy the required commutation relations in Eqs (7. (iii) ∆(0.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). φ− (y)] ≡ i∆+ (x − y).51) (7. = − (2π)3 2Ek (7.e.17) and (7.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. (7.48) (7. d3 k = (2π)3 2Ek = (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 (7.50) or as integrals over d4 k.18).52) The result for the invariant commutator function is [φ(x). In order to calculate [φ(x). (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . 2π (7. φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). (7.53) t=0 . x) = 0 and hence ∆(x) = 0 for x2 < 0.49) (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. i. The last item to be checked for the scalar ﬁeld are the causality condition. d3 k ei k·(x−y) = −i∆+ (y − x). a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (2π)4 (7.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . φ(y)] consider µν [φ+ (x). φ+ (y)] d3 k e−i k·(x−y) .55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .56) = −δ 3 (x).

b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). = ∂{µ φ∗ ∂ν} φ − L gµν . (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .62) (7.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). (2π)3 2Ek (7. a† (k ′ )] = [b(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) . 2 From the lagrangian density (7.e. extending it with the charge operator and the introduction of particle and antiparticle operators.Quantization of ﬁelds 60 7. The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) .64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x).63) i.60) = i φ∗ ∂µ φ.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ. The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (2π)3 2Ek (7. which is equivalent to the relations (only nonzero ones) [a(k). ↔ (7.57) (7.65) 7. (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). (2π)3 2Ek (7.59) (7.61) (7.66) . φ† (y)] = i∆(x − y). particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. we will consider it as a repetition of the quantization procedure. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).

(7. s)¯(k.74) (7.69) (7.70) where the last line is obtained by using the Dirac equation.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.73) (7. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.e. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . s)v(k. The solution is the introduction of anticommutation relations. s). s)ei k·x + d(k. † {ψi (x).71) (7. interchanging creation and annihilation operators. u v (2π)3 2Ek (7.77) Note that achieving normal ordering. d† (k ′ .67) i ψ / ψ − M ψψ gµν . jµ Θµν = = ψγµ ψ. (7. i.75) (7. s)¯(k. s)u(k. s) e−i k·(x−y) j s † + vi (k.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). s)u† (k. s). s)ei k·x . s) ei k·(x−y) x0 =y 0 . (2π)3 2Ek d3 k b† (k. {b(k. (2π)3 2Ek (7. b† (k ′ . then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . s)e−i k·x + d† (k. s)vj (k. ↔ i ψγµ ∂ν ψ − 2 (7.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : .78) (7. s′ )} = {d(k. The quantized ﬁelds are written ψ(x) = s d3 k b(k.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. ∂ 2 ↔ (7. s)e−i k·x . . (2π)3 2Ek (7.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained.

6 The electromagnetic ﬁeld 1 L = − Fµν F µν . −λ(∂ρ Aρ ). ∂ (7. (7.86) (7. but has the problem of being noncovariant. we would have obtained [ψi (x).84) (7.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. 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). (7. as the vanishing of Π0 induces a constraint. 4 2 This gives the equations of motion discussed before. ˙ δ Ai (7.88) . 7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.80) 3 = δ 3 (x − y) δij .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.85) which reﬂects the gauge freedom.82) where i∆1 = i∆+ −i∆− .81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. 4 Π0 Πi δL = 0. For the scalar combination {ψi (x). 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).83) From the lagrangian density the canonical momenta are = = (7. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . ∂ (7. 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. ˙ δ A0 δL = F i0 = E i . ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. ψ j (y)} at arbitrary times one has {ψi (x). it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. one has † {ψi (x).

94) Choosing k µ = (|k 3 |. It gives 3 k µ ǫ(λ) (k)c(k. (7. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). 6.e. λ). a longitudinal photon and two transverse photons. (7. [ℓi . ′ (7. (7. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. pj ]P . µ λ=0 (7.89) with four independent vectors ǫµ . i. where Q is the conserved quantity in Eq. j = 1. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). pj ]P .Quantization of ﬁelds 63 for physical states |A and |B . which reproduce the Hamilton equations. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). c† (k ′ . and [ℓi .92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators.11. Πν (y)] = 0.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). containing a time-like photon. µ λ=0 (7. 0. rj ]P . (d) Use Poisson brackets to calculate q and p. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. λ)c(k.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. These problems are solved by the Lorentz constraint between physical states given above. 7.3) satisfy the conditions of a Lie algebra. and are equivalent with [c(k. The canonical equal time commutation relations are [Aµ (x).95) Exercises Exercise 7. Aν (y)] = [Πµ (x). λ)e−i k·x + c† (k.91) (7. 0. q) that dO/dλ = [O. 0) . λ)ei k·x .e. 2. Q]P . λ) − c† (k. i. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. 0)c(k. (b) Proof for a quantity O(p. ˙ ˙ . In fact the commutation relations imply ˙ [Aµ (x).93) which does exhibit problems with the time-like photon. what are [ri . λ)|A = 0. for which it is suﬃcient that ∂µ Aµ |A = 0. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. 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.

Give also the expression for ∆C . (e) Show that ∆R (x) = 0 if x0 < 0. we can write for ∆R an integral where the k 0 integration remains along the real axis. (2 + M 2 ) ∆hom (x) = 0.Quantization of ﬁelds 64 Exercise 7. (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. (b) Check the above commutation relation [φ(x). Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7.2 Prove that the equal time commutation relations between quantum ﬁelds φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ ) Exercise 7. (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.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. f (Λk) = f (k) for Lorentz transformations Λ. ∆hom (Λx) = ∆hom (x) if f is an invariant function. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration). e−i a requires [φ(x). (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. Pµ ] = i∂µ φ(x). It is invariant.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. . (2 + M 2 ) ∆inhom (x) = −δ 4 (x).

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

1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x). T and C for classical quantities is shown in the table 1.2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. (iγ µ ∂µ − M ) ψ(x) = 0.1 Parity xµ = (t. writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). ˜ (8. The consequences of P. Both ψ p and ψ satisfy the Dirac equation. r) −→ xµ ≡ xµ = (t. x (8. and C quantity t r xµ E p pµ L s ˆ λ = s·p P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ T -t r −˜µ x E -p pµ ˜ -L -s λ C t r xµ E p pµ L s λ 66 .Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. Table 8. We can determine A. −r). parity (P). starting with the Dirac equation for ψ(x).1: The behavior of classical quantities under P. T. time reversal (T) and charge conjugation (C). 8.

λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.10) i. λ)v(k. the above operation reverses the sign of λ). −λ). one can also consider the scalar ﬁeld and vector ﬁelds. λ)γ0 v(k. −λ). 8. if helicity λ is used instead of the z-component of the spin m. m). λ) e−ik·x − d† (k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x).12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. λ)u(k. λ)u(k. −λ)v(k.7) = λ d k x x ηP b(k.3) (8. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. .9) (8. iγ µ† ∂µ − M ψ(˜) = 0. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. x The latter behavior of the vector ﬁeld will be discussed further below. m). In the same way as the Fermion ﬁeld. m) = = ˜ u(k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.e. λ)v(k. λ)γ0 u(k. ˜ Pop d(k.5) (8. −λ)u(k. ˜ −v(k. λ) e−ik·˜ + d† (k. x (8. λ) eik·x .4) (8. −λ) e−ik·x − d† (k. γ x (8. op P (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. −λ) e−ik·x − d† (k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. (2π)3 2Ek (8. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). λ) P −1 = −η ∗ d(k.11) (8. (i˜ µ ∂µ − M ) ψ(˜) = 0. λ) Pop = ηP b(k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). m) γ0 v(k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k.6) (check this for the standard representation.

One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . λ) e−ik·x + b† (k. 8.21) where A is a 4 × 4 matrix acting in the spinor space. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . λ). x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0.14) (8. x x (8. . antilinear operator is anti-unitary if A† = A−1 . x (8. The (time-reversed) Dirac equation becomes. λ)u(k. λ)v(k. The same relations hold for helicity states. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. λ)C uT (k. x (8. r) −→ −˜µ ≡ −xµ = (−t. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. m) ¯ C v T (k. λ) −1 Cop = ∗ ηC b(k. m). (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. λ) eik·x . λ)C v T (k. i. m) ¯ = = v(k.Discrete symmetries 68 the latter being also a solution of the Dirac equation. it is antilinear1 . (2π)3 2Ek (8.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .17) 3 = λ This shows that −1 Cop b(k. Norm conservation requires Top to be anti-unitary2 .19) Cop d(k.16) d k ηC d(k. (8. r). Top |φ = φt | = (|φt )∗ . Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. (8. x iγ µT ∂µ − M ψ(−˜) = 0. λ) e−ik·x + b† (k. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0.3 Time reversal xµ = (t. As time reversal will transform ’bra’ into ’ket’.e. we start with the complex conjugated x Dirac equation for ψ. λ) Cop = ηC d(k. λ). as discussed in section 4). u(k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.18) (8.15) (where one must be aware of the choice of spinors made in the expansion. m).

λ)iγ5 Cu(k.24) (8.26) d k x x ηT b(k.28) (8. p. λ)u∗ (k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜).4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. p. λ)iγ5 Cv(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. a state |a. −p. The result is = = λ (8. op T (8. λ) eik·x + d† (k.25) (check this for the standard representation). λ)v ∗ (k. λ) i γ5 C v(k. λ) eik·x + d† (k. −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 .29) In table 2 the behavior of particle states under the various transformations has been summarized.34) . λ). λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k.Discrete symmetries 69 Table 8. −p. p. ˜ Top d(k. −p. λ)u∗ (k. λ| ¯ C |¯.32) (8. Since there are 16 independent 4 × 4 matrices.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ). p. λ| a. λ) eik·˜ + d† (k. −λ a T a.31) (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯). x (8. λ a P |a. λ). there are 16 independent of these bilinear combinations. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. ∗ ˜ v (k. λ |¯. λ a |a. λ) eik·x + d† (k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. λ) Top = ηT b(k. −p.30) (8. λ) T −1 = η ∗ d(k. λ)v ∗ (k. λ)v ∗ (k. λ) e−ik·x (2π)3 2Ek (8. λ). λ)u∗ (k. (8. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. ˜ = 0|b(k)|A 8. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.33) (8.

Note that the lagrangian density L (x) → L (−x) under Θ. The 16 combinations of Dirac matrices appearing above are linearly independent. and T P S(˜) x Vµ (˜) x Tµν (˜) x −Aµ (˜) x −P (˜) x C S(x) −V µ (x) −T µν (x) Aµ (x) P (x) T S(−˜) x Vµ (−˜) x −Tµν (−˜) x Aµ (−˜) x −P (−˜) x Θ = PCT S(−x) −V µ (−x) T µν (−x) −Aµ (−x) P (−x) S(x) V µ (x) T µν (x) Aµ (x) P (x) The matrix σ µν ≡ (i/2)[γ µ . 8. γ5 γ µ or γ5 ). σ µν .g. p)u(p) e−i (p−p )·x . (q 2 ). (8.39) where Γµ (p′ . γ ν ]. and T.36) (8.41) 2M 2M . 2M (8. F.5 Form factors Currents play an important role in ﬁeld theory. called form factors.. e. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). p′ |V µ (0)|p = u(p′ )γ µ u(p).37) (8. As an example consider the vector current for a point fermion. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. C. p′ |V µ (x)|p = u(p′ )Γµ (p′ . C. as well as under the combined operation Θ = PCT (see Table 3). s . momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). for the vector current V µ (x)..38) (8. ¯ In general the expectation value between momentum states can be more complicated.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. (8. 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 ). 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). V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x .35) The x-dependence can be accounted for straightforwardly using translation invariance.Discrete symmetries 70 Table 8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. p) can be built from any combination of Dirac matrices (1. s′ |V µ (x)|p.40) where the coeﬃcients are functions of invariants constructed out of the momenta. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . p′ . ¯ ′ ′ (8. In many applications the expectation values of currents are needed. (q 2 ). of which the expectation value between momentum eigenstates can be simply found. in this case only q 2 = (p′ − p)2 . For instance for nucleons one has Γµ (p′ . F. γ µ .

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

p = (E. 0. 0. (b) Show that in the interaction with the electromagnetic ﬁeld.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. σµν q ν F3 (q 2 ) 2M . L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M .Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . (b) Show that such a term is also not allowed by time-reversal symmetry. 0. 0.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. 4M 2 = F1 + F2 . p′ = (E. (c) Show that hermiticity of the current requires that F1 en F2 are real. |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 . GE GM = F1 − Q2 F2 . |q|). (c) Show that if a term of this form would exist. Exercise 8. 0. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). Exercise 8. 0. −|q|/2).

and the operators are time-dependent. t′ ≡ q ′ |q ′ .1 Time evolution as path integral U (t. t0 ) = H U (t. t ′ |q . t′ . |ψH (t) = |ψH . t0 ) AH (t0 ) U (t. in which the operators are time-independent. t ′ QH (t′ )|q ′ . |ψS (t) = U (t. (9.6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing.4) ∂ U (t. (9. AS (t) = AS and the states o are time dependent. AH (t) = U −1 (t. (9. t0 )|ψS (t0 ) . QH (t)|q.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. Consider the two (time-independent) Heisenberg states: |q. ≡ 0. (9.Chapter 9 Path integrals and quantum mechanics 9.5) (9.2) (ii) Heisenberg picture. t0 ). i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. [AH . in which the states are time-independent. t0 ) ∂t (9. t . H].7) 73 . t ≡ q|q.3) (9. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture.

B]. B]] + [[A.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . B] + . use the Campbell-Baker-Hausdorﬀ formula.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . qj )]) . tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj .14) . [A.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . t′ |q. The hamiltonian is an operator H = H(P. ˙ 2π 1 1 1 [A. tj+1 |qj . (9. 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′ |q. 2π (9. tj+1 |qj .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . . qj ).. . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . |q1 q1 |e−i H∆τ |q .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj .8) Choose t as the starting point with |q = |q. 2 12 12 (9. H(P.12) where the transformation between coordinate and momentum space involves q|p = ei p. which is a hamiltonian function in which the operators are replaced by real-numbered variables. 2π (9.15) 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 . B] + [A. ′ (9. with the correction1 being of order (∆τ )2 . 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 . t = q ′ |e−i H(t −t) |q .11) (9. Then qj+1 . eA eB = eC with C = A+B+ (9.9) Dividing the interval from t ≡ t0 to t′ ≡ tn into n pieces of length ∆τ and using completeness (at each time ti ) one writes q ′ .16) this. The purpose of this is to calculate the evolution for an inﬁnitesimal time interval.q . Q) expressed in terms of the operators P and Q. n |q (9.. . Combining the two terms gives qj+1 . (9. QS |q ′ ≡ q ′ |q ′ . .

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

26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.25) (see Exercises).24) (9.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. .22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration. (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K. Consider the gaussian functional as an example.23) Having deﬁned the Gaussian integral.27) (9. Note that procedure (i) is an example of the more general procedure under (ii). 1 = √ det K 1 Dα Dα∗ exp (−α∗ K α) = det K 1 Dα exp − α K α 2 (α real-valued) (α complex-valued) (9. 2 (9.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m. 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. Dα F [α] = Dα F [α + β]. y) = ∗ m.21) again a multidimensional integral. with Dα F [α] = n dαn N F (αn ) (9. A useful property of functional integration is the translation invariance. N (9.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. (9. α(x) = n αn fn and K(x.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.

Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . e.g.e. In principle the deﬁnitions of functionals is the same. In the expansion of the exponential (from second to third line). θ2 = 0.37) 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. dx dy θ∗ (x) K(x. i.34) (9. n θn fn (x) dx θ∗ (x) θ(x) .29) (9. products of the same pair appear. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1.32) (9.33) (9.31) δ δ α = dy α(y) = dy δ(x − y) = 1. δα(x) or in discretized form δ 1 ∂ ∂ with . (9. however. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). δα(x) δα(x) δ αKβ = dy K(x. (9. θη = −ηθ. αn = δmn . It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. α(y) = δ(x − y). the Gaussian-type functionals. Examples of functional diﬀerentiation are (9. δα(x) (9. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .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. (9. which vanish. y) θ(y) .30) (9. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 .35) For applications to fermion ﬁelds. F [θ.38) . αy = δxy . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .

Path integrals and quantum mechanics

78

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

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

= δ(x − y),

(9.39)

(9.40) (9.41)

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

9.3

Time ordered products of operators and path integrals

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

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

q ′ , t′ |q, t =

Dq exp i

t

dτ L(q, q) , ˙

(9.44)

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

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

t′

Dq exp i

t i 2

dτ Leﬀ (q, q) , ˙

(9.45)

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

i

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

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

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

t′ t′ t

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i

t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

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

t t′

(9.49)

dτ L(q, q) . ˙

(9.50)

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

9.4

An application: time-dependent perturbation theory

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

′

(9.51) (9.52) (9.53)

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

t′ ∞ n=0 t′

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),

t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

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

**Path integrals and quantum mechanics
**

t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

′

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

t′

U

(n)

(t , t) = (−i)

t

′

dτ HI (τ ) U (n−1) (τ, t)

t′ τ1 τn−1

= (−i)n

t

dτ1

t t′ t′

dτ2 . . .

t t′

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

t

=

(−i)n n!

dτ1

t t

dτ2 . . .

(9.60)

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

t′ t′

+

1 2

dτ2

t τ2

dτ1 HI (τ1 ) HI (τ2 ),

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

t′ t′

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )

t′

+

1 2

t′

dτ2

t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

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

t′ t′

dτ1

t t

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

the desired result.

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

∞ t′ t′ t′

dτ1

t t′ t

dτ2 . . .

t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .

t

t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t|Q. q) + J(τ )q] . i. t q. t is related to the Schr¨dinger state |q by |q. t′ |q. Before and after these processes there is only the vacuum or ground-state |0 .5 The generating functional for time ordered products By introducing a source-term.Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. harmonic oscillator.62) p Dq D exp i 2π t dτ [pq − H(p. t|n = q|n e−iEn t with q|n the time-independent wave function. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. The Heisenberg state |q. q)] .65) J dq ′ dq Q′ . which in turn are embedded between an early time T and a future time T ′ . L(q.s. ˙ J (9.66) . a set |n of physical eigenstates. T < t < t′ < T ′ . for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. t|n n|Q. t′ |q. t Q′ . q) → L(q. q) + J(t) · q. φn (q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). for the physical states o q. ˙ ˙ physically pictured as. say.e. T with q. t) φ∗ (Q) e+i En T . Consider. making the electron) and the absorption of an electron (think of a detector). furthermore.e.63) for plane waves. 9. t′ q ′ . q)] ˙ (9. t |q. Considering the source to be present between times t and t′ .1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. n (9. T ′ |q ′ . T = n q. t = = ′ ′ V = q . q . T = n J t′ = = Dq exp i t dτ [L(q. the creation of an electron (think of a radio-tube.64) (9. T . t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. t|Q. for g. T ′ |Q. it is possible to switch on an interaction. t = ei Ht |q . one has q ′ . i.

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

+ V (q). . ˜ dt (9. (9. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. . . i dq . y) = ∗ m. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K.g.72) (9.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions.and Grassmann-valued functions. complex. ˜ ˜ ZE [J] δJ(t1 ) .77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. τ dq ˜ dt ≡ −L q. δ n Z[J] 1 Z[J] δJ(t1 ) . dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q).n fm (x) Kmn fn (y).Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. when L q. ensuring convergence for the functional integral ZE [J]. As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. where Kp are the eigenvalues of the matrix Kmn in K(x. .2 Check the examples of functional derivation for real-. δJ(tn ) J = 0 ˜ t = it (9.73) d˜ LE q. 1 = √ . det K Exercise 9.76) which is a positive deﬁnite quantity. The diﬀerentiations in Z[J] and ZE [J] are related. . (9. det K 1 . ZE [J] = = where LE q. for Grassmann valued functions.75) = − LE q. δJ(tn ) = (i)n J=0 1 δ n ZE [J] .74) − V (q) 2 (9. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . e. Exercises Exercise 9. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp .

det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω .80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . z) = δ(x − z). where dy K(x.79) (9.78) (9. y) K −1 (y. 2 (9. .Path integrals and quantum mechanics 84 Exercise 9.

x′ . i. . .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.4) J=0 . (10. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . . x. . Π) (10. φ(x.3) The Euclidean formulation is as before. . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. xn ). .e.Chapter 10 Feynman diagrams for scattering amplitudes 10. . δJ(xn ) 85 ≡ G(n) (x1 .2) = ∞ −∞ d4 xE [LE (φ.1) Dφ exp i σ d4 x L (x) and T′ Z[J] = lim T → i∞ T ′ → −i∞ Dφ exp − Dφ exp i T d4 x [L (φ. 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 ) . φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . . ∂µ φ) − Jφ] (10. t′ |φ. ∂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 . with Z0 [0] = 1.13) . . . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. (2π)4 (k 0 + iǫ)2 − E 2 (10.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. (10. .10) J=0 In order to determine ∆F . Furthermore.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . J(xn ). As discussed in the previous section (exercise 9.e.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x .3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . . . . (10. . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . . xn ) J(x1 ) . . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . 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. . . 2 2 → (10. (10. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. d4 xn G(n) (x1 .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. xn ) = x2 xn .9) i. .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y).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).

22) (10.20) = exp − 2 where or (see ﬁg. 4 k 2 − M 2 + iǫ (2π) (10.4). .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (2π)4 k 2 − M 2 (10.14) They are solutions of the inhomogeneous equation.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x .Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. the advanced Green’s function ∆A (x) = satisfying ∆A (x. (10. 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). (2π)4 k 2 − M 2 (10. ∆F = ∆C . The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. µ 1 1 • Firstly.18) Thus we see various solutions. 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 . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).16) C where C is a closed contour in the complex k 0 -plane. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. ∆(x) = − d4 k e−i k·x . t) = 0 for t > 0.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . (10.17) d4 k e−i k·x . In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. 10. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x).e. (2π)4 (k 0 − iǫ)2 − E 2 (10. t) = 0 for t < 0. φ+ (0)] = (2π)3 2E are also shown in exercise (7.15) e−i k·x d4 k . i.

the same result as above. . 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. we have G0 (x1 . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . . we use the fact that i∆F (x − y) = G(2) (x − y) = 0|T φ(x)φ(y)|0 to calculate ∆F via the time-ordered product.Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F .25) x1 x2 x3 = x4 + + . = θ(x0 − y 0 ) 0|[φ+ (x). φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x).3) i ∆F (x − y) = 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0 = i ∆C (x − y). 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. We ﬁnd (see also section 7. x3 .24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. because this is the only quantity entering Z0 [J].

27) When interactions are present. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . x2 ) = i ∆F (x1 − x2 ) = G(4. the generating functional can be written.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.g ) (x1 .g 0 ) = G0 = 1. (10.30) and in zeroth order G(0. (2) (0) (10. (10. i. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. x3 . x2 ) = G0 (x1 .Feynman diagrams for scattering amplitudes 89 10. x3 . x4 ) = G0 (x1 . (10. x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.29) This expression will be the one from which Feynman rules will be derived.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 (10.31) G(2. x2 .35) z .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) (10.28) 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 . with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].32) (10.g ) (x1 . Z[J] = = Dφ exp i exp i d4 z LI (10. L (φ) = L0 (φ) + LI (φ).34) Introducing a vertex point to which four propagators are connected. x2 .

i. . δJ(xn ) . G(n) (x1 . .37) G(2. did not contain parts that could be written as products of simpler Green’s functions. xn ) J(x1 ) . . 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]. (10.41) in terms of socalled connected Green’s functions. (10. (10.39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. . . . In the free case we have seen that the exponent of Z[J] contained all essential information. .g ) (x1 . .40) in−1 n! d4 x1 . . d4 xn G(n) (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].Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . Z[J] = n in n! d4 x1 .g 1 ) = 1 8 z 1 8 . .38) The result for the 4-points Green’s function is left as an exercise. xn ) J(x1 ) . . . . . . . (10. . which also was the only Green’s function for which the diagram was connected. . . Diagrammatically this exponent only contained the two-point function.36) J J from which one obtains G(0.42) x1 G(n) (x1 . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . J(xn ) c (10. . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . . J(xn ). J=0 (10. . . This remains also true for the interacting case. (10. d4 xn G(n) (x1 . . . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 .e.

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

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

a) = S. 2 ∂µ ∂ µ + M0 φ(x) = J(x). out|φout = β. φin (y)] = iZ ∆(x − y). this can actually not be used as it would imply [φ(x). in|φin S → φin S = S φout (x).54) . Proof: β. e. S is invariant under e Poincar´ transformations: U (Λ. The relation between S-matrix and in. (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. 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. (10.and out-ﬁelds is: φin (x) = S φout (x)S −1 .49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions.52) where Z is a constant. as it stands. the source term Jφ in the lagrangian density is treated in the interaction picture. (10. The convergence therefore must be weakened to t→−∞ √ (10. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10. (2π)3 2E (10.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . z) φ(z).g.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 ). z) φ(z) d4 y d4 z ∆A (x − y) K(y. √ Although the above.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. 9. implies the strong (operator) convergence φ(x) → Zφin (x). a)SU −1 (Λ. The relation between S and Z[J] To establish this relation.51) (10. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. i.48) The fact that φin and φout satisfy the homogeneous Klein-Gordon equation with the physical mass M implies that they create particles and antiparticles as discussed. φ(y)] = Z [φin (x).e. a causality condition that can be proven to imply the absence of interactions. which in a later stage (renormalization) will become more important. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . β. 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).

e = i.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits.58) In order to ﬁnd a solution to this equation note that.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). eB ] = [A. B] and [A. φ(x). express it in terms of advanced and retarded Green’s functions.59) (10. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. B] (for the case that [A. B + C]e e . z) d4 y d4 z ∆(x − y)K(y. z) δU [J] .e.62) where F [J] is some (arbitrary) functional. z) . C] are c-numbers. φ(x). Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). just as we did for φ(x). δJ(z) Taking the diﬀerence between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.57) i [φin (x). e e ] = [A. (10. which is equal to the expression e in which the creation operators are placed left of annihilation operators. y) δ δJ(y) : F [J]. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. we can.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). δJ(z) (10. δU [J] δJ(x) = = √ i Z U [J] φin (x) − √ i Z φout (x) U [J] − d4 y d4 z ∆R (x − y) K(y. δJ(z) (10. t i Ut∞ [J] φ(x)U−∞ [J] (10. S U [J]] = √ Z δ S U [J].Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. B]eB . z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y.63) . [A.61) e φ+ f where the normal ordered expression : e φ f : is used. one has [A. provided [A. (10. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. φ(y)] f (y) e e . Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. φ− f (10. B] is a c-number).

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

= −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). but overall momentum conservation at a vertex is understood. Diagrammar) in the case of the interaction terms g f (10. If problems arise go back to the deﬁning expression for the generating function in 10. the factor 2 corresponds to the two-points Green’s function having two identical ends). 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines ﬁxed).69) LI (φ) = − φ3 − φ4 . to be precise iLI vertices in momentum space are introduced. ’t Hooft and M. Veltman. Note that in calculating amputated Green’s functions external lines are neglected. 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. For the interaction terms in the lagrangian.27. = −i g .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 2.27. or calculating full Green’s functions external lines are treated as propagators.66 and 10. which cancels the factor 1/2 in the quadratic piece. For the symmetry factor consider the examples (given in G. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically diﬀerent connected diagrams. At these vertices each line can be assigned a momentum. vertices and external particle wave functions in that diagram. 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.

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

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

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

76) 1 . s′ )γµ u(k. s4 )γµ u(p. s)¯(k.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s3 )(−ie)γ µ u(k. t2 (10. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. s4 )(−ie)γ ν u(p. To lowest order (∝ α = e2 /4π) only one diagram contributes.77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s)γν u(k ′ . s1 ) ¯ −igµν u(p′ .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s1 ) u e2 u(p′ .75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. The 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 ′ .79) . s3 )γ µ u(k. where u ¯ L(m) µν = = 1 2 u(k ′ .s2 . s′ ) ¯ u s.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .s4 e4 (m) µν (M) L L . s2 )|2 ¯ q2 (10.s3 . s2 ). 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 .Feynman diagrams for scattering amplitudes 100 10. ¯ q2 (10.s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10.

1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 .s4 = 1 2s × v (k ′ .s3 . (4) (4) . 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. Exercises Exercise 10. s1 )γν v(k ′ . . the masses. s1 )|2 ¯ s 1 . (c) Do the same for the connected Green function Gc . The diagram.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ .s2 3 .s2 . s2 (10. . s2 )γµ u(k. s3 )γ µ v(p′ . . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. s3 )γ µ v(p′ . Similarly. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s4 ) u e2 v (k ′ . s3 ) ¯ ¯ 1 . s4 )γ ν u(p. s2 )γµ u(k. s1 ) u(k. This is known as crossing symmetry. x4 ) for the interacting case in φ4 -theory to ﬁrst order in the coupling constant g as obtained from the full expression for Z[J] in section 10. s2 ) ¯ ¯ 1 2s u(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 .2. s4 ) v (p′ .Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. To lowest order (∝ α = e2 /4π) only one diagram contributes.

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

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

12) pb = (mb . m2 . 2mb (11.15) Explicit calculation of the reduced two-body phase space element gives dR(s. 0). Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . p1 . ptrf ). m2 ) a b . Also in this case one can express the energy and momentum in the invariants.14) One can.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . p2 ) = = where λ12 denotes λ(s. p1 . a (11.9) (11.8) (11. 4π s 4π 8π s 4π (11. m2 . which in the speciﬁc case a b also can be expressed as λ(s.13) (11. (11.11) (11. 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 ).10) The function λ(s. m2 . 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = . mb and s). 4s (11. m2 . 2 s s − m2 + m2 a b √ . 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. for instance. 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. m2 . In that case trf pa = (Ea . trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 2mb λ(s. m2 ) is a function symmetric in its three arguments. m2 ) = 4(pa · pb )2 − 4p2 p2 . = 2 s λ(s. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest.16) . m2 ) a b .

In inclusive measurements no particles are detected in the ﬁnal state. At high energies. in exclusive measurements all particles are detected. Instead of the scattering angle. there is usually a preferred direction. Writing E = mT cosh y. in which one particle is detected. Consider the 1-particle inclusive case.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance).27) with m2 = m2 − p2 = m2 + p2 + p2 .25) dR(s. pT = (px . = (11. usually many particles are produced. py ).19) (11. tmax min λcd /4s. a c with q = λab /4s and q ′ = being 0 or 180 degrees.17) (11. The variable s is always larger than a c b d the minimal value (ma + mb )2 . 11. . however.. E .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. 2s (11.18) (11. The various possibilities are referred to as diﬀerent reaction channels.. 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 .21) (11. Consider the process a + b → c + d.4). Of course there are not only 2-particle channels. (11. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . pd ) = 4π s 4π 8π s 4π dt dt √ √ . 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 .20) → π0 + n → π+ + π− + n → . A speciﬁc reaction channel starts contributing at the threshold value (Eq.22) (11. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . pc . s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. pz = mT sinh y. An often used set of invariants are the Mandelstam variables. H1 + H2 → h + X. The initial state. usually is a 2-particle state. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. Examples appear in → π− + p (11.

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.28) which (only involving angles) is easier to determine. t = (p1 − p3 )2 = (p2 − p4 )2 . These variables are precisely the Mandelstam variables for the s-channel process (ab → cd).2 Crossing symmetry In the previous chapter. 11. t. c bd Mad→c¯(su = u. which means that rapidity distributions dN/dy maintain their shape. ut = u) = Mab→cd (s. uu = s) = Mab→cd (s.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. t. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. tt = s. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. ¯ tt = (pa − p¯)2 = s. ¯ b (11. uu = (pa − p¯)2 = s. u = (p1 − p4 )2 = (p2 − p3 )2 . The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . b ut = (pa − pd )2 = u. This is known as crossing symmetry. t-channel and u-channel processes respectively.30) (11. u). ¯ tu = (pa − pc )2 = t.31) . u). 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. η = − ln(tan(θ/2)) = tanh−1 (cos θ).29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. tu = s. (11. (11. t. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t.

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

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

(11. . The result is N = |Mf i |2 dR(s. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E.e. q f ). (11. . √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). pn ). θcm ) √ q 8π s 2 2 (11.43) dΩcm = π M (s.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. see 11. . . 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 . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . 16π 2 s (11. (11.44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. (11. i. q n ). 2 λab and for instance for two particles dσ = q ′ M (s. . 11. θ) = −8π s ℓ (in analogy with the expansion for f (E. 8π s 2π 8π s . ˆ ˆ Inserted in the unitarity condition for M . . pn ) |M |2 q 32π 2 m2 dΩ |M |2 . for which one has dσ/dΩ = |f (E. . θ) or M (q i . for which the √ amplitudes squared have been calculated in the previous chapter. implies for the scattering matrix M .47) M (s. θ). (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude.41) (11.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. pn ). θ) in quantum mechanics. p1 . . . (11.Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous).25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. θ)|2 .40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . note the sign and cos θ = qi · qf ). t) λab 4π dt. . . p1 . q n ) q n qn √ Mni (q i . respectively. p1 .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM.

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

pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. For instance the tree-level calculation of Feynman diagrams at order e2 was real. i. the time-evolution operator. 11.57) This shows that Γ is the width of the resonance.e. 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.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. The product of amplitudes is of order e4 . For an unstable particle one expects that U (t. . s − M 2 + iM Γ (11.56) (again disregarding spin). 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 . The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . however. i. but still real. . To check unitarity we need the amplitude at order e4 . 0)|2 = e−Γt . such that |U (t. ∝ U (t. in which case. 0). It is important to note that the physical width of a particle is the imaginary part of the two-point vertex at s = M 2 . (q Mℓ )ij (s) = −M Γi Γj . . σel = σinel . 0) ∝ e−i(E−iΓ/2)t . it is straightforward to write down the partial wave amplitude. For the amplitude in a scattering process going through a resonance.55) (note that we have disregarded spin). The prescription for the pole structure. (11. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. The quantity Γ is precisely the width for unstable particles.58) . . which is given by a sum of the partial widths into the diﬀerent channels. This is what we will do to discuss decay widths. η = 0. We note that unitarity is generally broken in a ﬁnite order calculation.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. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption.

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

999 877 Γ and Γ(π − → e− νe ) = 0. cm |va − vb | in the center of mass system for two colliding particles a and b. Exercise 11. Compare the results with the experimental π − lifetime τ = 2. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. ρcm ρcm a b 1 cm . Calculate now fπ . 4. writing N (s) M1 (s) = 2 + iM Γ . Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. a quantity which we will ¯ ¯ encounter in the next chapter (section 12.3 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. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . Thus.8.2 Calculate the ﬂux factor.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Exercise 11.2).Scattering theory 113 Exercise 11.4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ .000 123 Γ. As discussed for unstable particles. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude.

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

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. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. should be independent of such transformations. However. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).15) ψ(x). the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . (12.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). ds which is solved by dψ(s) ds = dxµ . requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). -dx -dy µ µ dyµ µ dx For a vector.19) This gives rise to a (path dependent) phase in each point. i. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). which connects two identical vectors.16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx).e. = . (12. but expresses them with respect to diﬀerent bases. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. (12. The relation in Eq.16. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0.14) (12.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. i. ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). 12. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. (12.e. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. In principle such a phase in a given point is not observable.The standard model the basis for x G and x+dx 117 G. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x).

and equivalently Aµ can be considered as a pure gauge eﬀect. a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components.e.g. the groundstate itself appears degenerate. If it is zero one has dxµ Aµ (x) = 0. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect). φ1 φ . . If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. this means that there is more than one possibility for the groundstate.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. Nature will choose one. Dν ]. In this case there are many possible groundstates (determined by a ﬁxed direction in space). e. external magnetic ﬁelds. 12.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν . Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . As there can be one and only one groundstate.2. usually being (slightly) prejudiced by impurities. φ= 2 φ3 (12. 2 2 4 −V (φ) The potential V (φ) is shown in ﬁg. This characterizes spontaneous symmetry breaking. Nevertheless. i. Similarly as the deﬁnition of the covariant derivative the eﬀect is thus frame-independent and we have the transformation law Gµν → S(x)Gµν S −1 (x).2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. i>j which is rotationally invariant. 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. 12. we can disregard those ’perturbations’ and look at the ideal situation.The standard model 118 where Gµν = (i/g)[Dµ . in reality a not perfectly symmetric situation. Only if this quantity is nonzero a physically observable eﬀect of Aµ exists.

The situation now is the following.23) F +η the latter only forming a minimum for m2 < 0.22) F Realization of symmetries (model independent) In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented. i.The standard model 119 a constant ﬁeld (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2. The η massless particles are called Goldstone bosons. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. . etc.) terms constitute interaction terms. In this case the spectrum is described . so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. They are known as the Weyl mode or the Goldstone mode: Weyl mode. . (12. ϕc = 0|φ |0 = (12. . The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. (12.25) 2 2 2 Therefore there are 2 massless scalar particles. therefore a choice must be made.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). The classical groundstate appears degenerate. for the vacuum Qa |0 = 0.e. Any constant ﬁeld with ’length’ F is a possible groundstate. while the quadratic terms must correspond with non-negative masses. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . (12. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . has less symmetry. 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 . say 0 0 .24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . ϕ2 and η. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . It is only invariant under rotations around the 3-axis. In the case of degeneracy. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . In this situation one speaks of spontaneous symmetry breaking. 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. while the lagrangian including the nonzero groundstate expectation value chosen by nature. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic.

. . T = τ /2.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). . denoted |π a (k) . This means that they are operators that create states from the vacuum.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. Goldstone mode. (12. H] = 0. 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′ . a massless Goldstone boson for each ’broken’ generator. [Qa . it is easy to see that one gets ∂µ V µ = i ψ [M. i. a and a′ are degenerate states.The standard model 120 by degenerate representations of the symmetry group. T ] ψ. 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.. if the generators Qa generate a symmetry. ∂ ∂ (12.e. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space.e. Using the Dirac equation.32) (12. = ψ(i/ − M )ψ. Lx and Ly or instead of the latter two L+ and L− ) are used to label the multiplet members or transform them into one another.31) . a ∂ µ Jµ = 0. M = .28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. The generators Qa (in that case the rotation operators Lz . Neglecting at this point the local color symmetry. etc. 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. ψu mu md ψ = ψd . a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab .26) for all the states labeled by a corresponding to ’broken’ generators. Taking the derivative. there must be a nonzero expectation value 0|Jµ (x)|π (k) . (12. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . ψ −→ ei α·T ψ. Including a sum over the diﬀerent ﬂavors (up. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). .e. i. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). called ﬂavors. A bit more formal. down. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. π (12. and one must have mπa = 0. strange. i. V µ = ψγ µ T ψ.) one can write L = f ψ f (i/ − Mf )ψf . (12.

i. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. or.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. (12. however. This situation.e. 12. if in ﬂavor space M is proportional to the unit matrix. I. There exists another set of symmetry transformations. both are very small. Therefore the world of up and down quarks describing pions.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices.g. where the quark masses are not completely zero.34) Using the Dirac equation. From the number of broken generators it is clear that one expects three massless Goldstone bosons.35 is nonzero. parity minus). This symmetry.6 MeV/c2 ). stated equivalently. etc. T } γ5 ψ. i. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . but also a triplet (isospin 1) of pions. T ] = 0.3 MeV/c2 and Mn = 939. proton and neutron. but the fact that the symmetry is not perfect and the right hand side of Eq. (leaving out the ﬂavor structure) the same as ψγ5 ψ. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. (12.27.36) 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. From group theory (Schur’s theorem) one knows that the latter can only be true. the lagrangian density is invariant under left (L) and right (R) rotations independently. T γ5 = τ γ5 /2. chiral symmetry is not perfect. is what happens in the real world. it is easy to see that one gets ∂µ Aµ = i ψ {M. for which the ﬁeld (according to the discussion above) has the same behavior under parity. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. M = m · 1. In the real world. which is equivalent 2 to the V-A symmetry. .e. much smaller than any of the other mesons or baryons. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. 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.e. ψ −→ ei α·T γ5 ψ. including in Dirac space a γ5 matrix. which is approximately true for the up and down quarks.e.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . as the quantity ∂µ Aµ (x). The currents corresponding to this symmetry transformation are again found using Noether’s theorem. socalled axial vector transformations. behaves as a pseudoscalar particle (spin zero. while the groundstate is only invariant under isospin rotations (R = L). The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 .e. This combined symmetry is what one calls chiral symmetry. fπ = 0 remains. with almost degenerate masses (Mp = 938. Aµ = ψγ µ T γ5 ψ. i. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. parity would not play a role in the world. generated by TR/L = 1 (1 ± γ5 )T . etc. (12. ∂ ∂ (12. The chiral ﬁelds ψR and ψL are transformed into each other under parity. ψR/L = 1 (1 ± γ5 )ψ. How can we see this. is by nature not realized in the Weyl mode. e. 12. a doublet (isospin 1/2) of nucleons. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . Note that these transformations also work on the spinor indices. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved.

. in this case three-dimensional) representation reads (La )ij = −i ǫaij . like a photon. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). as in this case there are no massless Goldstone bosons. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). thus 9 independent degrees of freedom. . Initially there are 3 massless gauge ﬁelds (each. . .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 . Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. i.The standard model 122 12.40) a Dµ φ = ∂µ φ − ig Wµ La φ. We have the possibility to perform local gauge transformations. 0 ϕc = 0|φ |0 = 0 . One may wonder about the degrees of freedom. 0 0 0 = . two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . made into a gauge theory. 4 2 where Since the explicit (adjoint. again 9 degrees of freedom. 0 φ= (12. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. (12.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . (12. (12. After symmetry breaking the same number (as expected) comes out.39) (12. is made. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. having two independent spin components) and three scalar ﬁelds (one degree of freedom each). 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 + . Dµ φ = Gµν = ∂µ φ + g Wµ × φ.37) (12. The diﬀerence comes when we reparametrize the ﬁeld 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). The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle.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.e. The symmetry is broken in the same way as before and the same choice for the vacuum. but one has 1 massless gauge ﬁeld (2).42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). while the number of massive gauge bosons coincides with the number of ’broken’ generators.

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. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . etc. ψR/L = 2 (1 ± γ5 )ψ decouple.47) and YW is considered as an operator that generates a U (1)Y symmetry. 2 (12. neutrinos.48) (12. νµ and ντ ) and their antiparticles. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. that gives a good description of the physical world. R. 1 As they are massless. muon µ− or tau τ − ). (12. explicitly L −→ e−i β/2 L.The standard model 123 12. while the lefthanded particles form a doublet. i.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. 2 2 = ∂µ R + i g ′ Bµ R. the quarks (up.and righthanded particles.45) (12. electromagnetic and weak forces.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.e.50) (12.51) .46) −→ 3 One notes that the charges of the leptons can be obtained as Q = TW − 1/2 for lefthanded particles 3 and Q = TW − 1 for righthanded particles.) and the gauge bosons responsible for the strong. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . (12. The procedure. down. This is written as 3 Q = TW + YW . explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. = ∂µ L − (12. U(1)Y U(1)Y (12. ∂ ∂ ∂ (12. etc. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. the leptons (elektrons.44) which has an SU (2)W symmetry under transformations eiα·TW . however. muons. left. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. R −→ e−i β R. its corresponding neutrino (νe . Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. τ ).43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.).

57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1.59) .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. Putting this in leads to L (f ) =L (f 1) +L (f 2) . Using local gauge invariance θi for i = 1. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). L where L (h1) (h) =L (h1) +L (h2) . For the classical ﬁeld the choice θ4 = v is made. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. 4 4 In order to break the symmetry to the symmetry of the physical world. .53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. leading to the parametrization 1 0 φ(x) = √ (12. (12. the U (1)Q symmetry (generated by the charge operator). (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. −V (φ) L and (h2) = −Ge (LφR + Rφ L). h) (12.56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms. . + 8 (12. . (12. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. . . † i i g Wµ · τ − g ′ Bµ )φ.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 .58) (12. this leads to the lagrangian L (h1) .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .

2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2 (12. (12. .60) (12.61) (12. g ′ /g = tan θW .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 .64) (12. The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.69) . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ .67) The coupling of electrons or muons to their respective neutrinos. . (12. 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 .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . 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 + .66) From the coupling to the photon.63) (12. we can read oﬀ e = g sin θW = g ′ cos θW . 2θ 2θ 4 cos W cos W 0. 4 2 MW g2 v2 = . 2 Zµ (12.65) The weak mixing angle is related to the ratio of coupling constants.

5 MeV (exp.71) GeV −2 = = = = . (12. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. one has the relation g2 e2 1 GF √ = .80) First.166 4 × 10 0. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. (12. 91. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.e.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).74) (12.40 GeV. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. 3 = TW .77) (12.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.g.e. g2 MZ 2 2 (CV + CA ). g ′ and v determine a number of experimentally measurable quantities. (12.231 2. giving rise to a running coupling constant after renormalization of the ﬁeld theory. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . from the MW mass is about 250 GeV. i. Comparing this with the total width into (invisible!) channels.19 GeV. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. v (12. The coupling to the Higgs particle is weak as the value for v calculated e. 80. i. With the choosen vacuum expectation value for the Higgs ﬁeld. Finally we want to say something about the weak properties of the quarks. 2 2 2 (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 .70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.73) (12. me /v is extremely small.76) with CV CA 3 = TW − 2 sin2 θW Q. . −5 (12. (12. 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.72) (12. 1. muon or tau.

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

g. respectively).22 and η ≈ 0. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. d d (12. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks.e.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .227.83) where Vu and Vd are unitary matrices. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. The Higgs ﬁeld is still limited to one complex doublet. i. (12.34. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . u u where we include now the three lefthanded quark doublets in QL .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. e. the weak mass eigenstates are L (h2. ms and mb ). UR = uR cR tR . magnitudes of the entries in the CKM matrix are nicely represented parametrization 1 − 1 λ2 λ λ3 A(ρ − i η) 2 1 2 −λ 1− 2λ λ2 A V = 3 2 λ A(1 − ρ − i η) −λ A 1 12.82) space. . allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h .82 and ρ ≈ 0.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. each of these containing the three families. † Λu Λ† = Vu G2 Vu .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. One then reads oﬀ that starting with the family basis as deﬁned via the left doublets that the mass eigenstates (and states coupling to the Higgs ﬁeld) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . (12.86) the unitary CKM-matrix introduced above in an ad hoc way. A ≈ 0. 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 . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .

decoupling from all known interactions.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. 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. EL = Ve EL . UPMNS a parametrization that in principle also could have been used for quarks.89) with three righthanded neutrinos added to the previous case. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn .91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. there is no family mixing for massless neutrinos. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). e N L= L EL (12.93) 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 . NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. (12.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .e. (12.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. 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 . The mass ﬁelds ER † mass † mass = We ER .5×10−3 eV2 .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . one can write Λe = Ve Ge We and we ﬁnd † L (h2. As before.90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . and obtain L (h2. In this case. (12. i. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . EL = Ve† EL . The mixing is intriguingly close to the Harrison-Perkins-Scott tri-bimaximal 23 mixing matrix 0 0 2/3 1/3 0 2/3 1/3 0 1 1/2 − 1/2 − 1/3 UHPS = 0 1/3 − 1/2 . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. mµ and mτ . (12. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . since the weak current is the only place where they show up.92) .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . we ﬁnd massless neutrinos. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.ℓ) = −LφΛe ER − ER Λ† φ† L.

c. a unitary matrix relating them to the weak eigenstates. 2 R ψ = nR + nL . The way to circumvent this is to introduce as in the previous section righthanded neutrinos. Absorption of phases in the states is not possible for Majorana neutrinos. 1 L Υ ′ = nR + nc . the massless and massive Majorana neutrinos. with for the righthanded sector a mass term MR . − MD MR nR 2 (12.ν = − MR NR NR + MR NR NR .ν =− containing three (CP-violating) phases (α1 . coupled by a Dirac mass term.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. 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. one must for the neutrinos as well as charged leptons.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 .and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. the left. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. as above. For the neutrino sector. For these light Majorana neutrinos one has.and righthanded species then just form a Dirac fermion.98) (12.94) M L NL NL + M L NL NL .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. Thus (disregarding family structure) one has two Majorana neutrinos. For the leptons.95) 2 In order to have more than a completely decoupled sector. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. 12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass.99) . Thus 2 Υ ′ = nc + nL . The mass matrix can be written as ML |MD | eiφ M = (12. We use here the primes starting with the weak eigenstates. one being massive. 1 c ∗ c L mass. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). however. This is called the see-saw mechanism (outlined below). Actually.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. are equivalent to two decoupled Majorana neutrinos (see below). 2 although this option is not attractive as it violates the electroweak symmetry. 0 0 1 L mass. is to add in Eq. (12. couple the right.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. hence the mixing matrix becomes iα /2 0 0 e 1 (12. 1 ∗ c c (12. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. α2 and δ). (12.

H] = 0. Prove that if the generators Qa generate a symmetry. i. Exercise 12.1 Consider the case of the Weyl mode for symmetries. 2 with real masses Mi .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. T } γ5 ψ. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. The diagonalization is analogous to what was done for the Λ-matrices and one ﬁnds U M U T = M0 with a (unitary) matrix U . Υ2R nR (12. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . L Υ2L nR Υ1 and Υ2 . i ψ {M.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates.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)) . Exercises Exercise 12. (12.105) for i = 1.101) 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 . i. (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′ . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . .e.The standard model 131 taking ML and MR real and non-negative. related via c Υ1R = U nL . This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families). [Qa . (12. 2 U (M M † ) U † = M0 . a and a′ are degenerate states.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 . Exercise 12. T ] ψ.e.

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

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

relativistic quantum mechanics

relativistic quantum mechanics

- Book 7
- quantum field theory
- M. Kozlov, E. Levin, V. Khachatryan and J. Miller-The BFKL Pomeron Calculus in zero transverse dimensions
- Quantum Field Theory 2013
- 6.17.docx
- einstein general relativity
- vak
- Chapter 14- Iso, Geno and Hyper-Grand-Unifications and Iso,Geno, Hyper-Cosmologies:14.1- Iso, Geno and Hyper-Grand-Unification
- FunkyQuantumConcepts.pdf
- Chapter 10- Grand-Unification and Cosmology
- Monopole condensation in two-ﬂavour Adjoint QCD
- 9806024v1[1]
- Quantum Mechanics - Concepts and Applications - T. Biswas
- RevModPhys.66.129
- Notes 2003
- Paper on the Aokiphase 2008
- 2Q_Fradkin
- 2nd_lecture1
- UCB 221A - Postulates of Quantum Mechanics
- PhysRevLett.55.2850
- Proc_vietnam_2007
- Hydrogen Atom 2D
- Probability Density in the Complex Plane
- Infrared Behavior of Scalar Condensates in Effective Holographic Theories
- A Report on How Friction Factor is Obtained
- V.A.Berezin, A.M.Boyarsky and A.Yu.Neronov- Towards the mass spectrum of quantum black holes and wormholes
- lotka ecuaciones 2
- LC_lock

Are you sure?

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

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

Get the full title to continue

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

scribd