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

1 2.2 6.2 3. Addison-Wesly. Vol.1 2 2 3. 3. 2. 5. Vol.2 1.3 2.3 2.2 2.-B.3 2. The quantum theory of ﬁelds.2 6.2 2.E. S. 2007.3 References As most directly related books to these notes.3 3. 1985. C.4 2.3.5 2.2 3.5 2.4 3. Corresponding chapters in books of Ryder.1 3.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.4 4.2. I refer to the book of Srednicki [1] and Ryder [2].3 3. Itzykson and J.3 2. The notes contain all essential information. M. Cambridge University Press.2 3.1 2. Cambridge University Press.4 6. Peskin and D.3 6. 1980. I: Foundations. but are rather compact. Peskin & Schroeder and Srednicki. M.3 2. M. 3. L. 1995. Ryder.1. An introduction to Quantum Field Theory.4 3.1 1.1 3. Cambridge University Press. Zuber. Cambridge University Press.5 4.3 4. 1994. 1995. Weinberg. 2. 1.2 36 22 . 3.3 3.2 3. Veltman.3 3.V. 1996. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. Cambridge University Press. McGraw-Hill. Srednicki.8 3. II: Modern Applications.5 5. 6.1 4.2 4.7 2. Quantum Field Theory. Schroeder. 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).4 2.1 5. Quantum Field Theory.H.2 3.4 3. 4.2 2.1 6.3 3. Quantum Field Theory. These notes Section 1. Diagrammatica.

4.2 8.1 20.1 20.4 2.3.3 8.1.3 6 8 9 Srednicki 3 3 3 37 5.8 5 4.8 6.2 4.7 4.4 9.4 4.6 3. 4. 6.1 9.4 8.4 12.5 .6 6.7 5.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2. 4. 20.6 4. 4.3 10. 9.5 7.5 6 7 6 6.6 10.1 11.5 9. 6.3 7.3 6.1 5.7.5 7.5 6.2.4.5.1 8.3 20.1 12.2 9.6 3.3.3. 9.4 10.2 11.2. 2.5 4.3 9.2 12. 6.3 12.3 3.4 3.3 11.1 20. 6.6 9.4.6 3.2 8.2 10.2.5 4. 2. 8.1 10.4 These notes Section 7.2 5.5 11.2.10 8.1.5 9.4 7.4 12.5 2.1.4.1 9. 6.3 8.2 7.3 7.6 8.1 7.3 2.

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.

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

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.

the points lie on the lightcone and they can be connected by a light signal. The real. x3 ). When 3-dimensional space is considered as part of Minkowski space. and are given by (x0 . x1 . Unlike in Euclidean space. Many other four vectors and tensors transforming like T ′ µ1 . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero)..1. x1 . 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. denoted as V ′µ = Λµν V ν . x) = (x0 .3) with basis nµ (µ = ˆ 0. The length (squared) of a vector is obtained from the scalar ˆ product. In Minkowski space.2. x2 .25) Within Minkowski space the real. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .Introduction 5 that can be used in the cross product of two vectors z = x × y. (1. −x). one must distinguish tensors with upper or lower indices and one can have mixed tensors. 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. • ds2 = 0 (lightlike or null intervals). Vectors are denoted x = xµ nµ . x2 = x · x = xµ xν nµ · nν = xµ xν gµν . One can distinguish: • ds2 > 0 (timelike intervals). x2 . These transformations do change the components of a vector. independent of a coordinate system. . Because of the diﬀerent signs occurring in gµν . however. • ds2 < 0 (spacelike intervals).26) (1. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations.µn = Λµ1ν1 . Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. are . δim δjm δkm δin δjn δkn . xµ = gµν xν .21) (1. The lower indices are constructed in the following way. In special relativity we start with a four-dimensional real vector space E(1.23) δjm δkn − δjn δkm . (1. x3 ) = (t. Relations relating tensor expressions.. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. where the coordinate t = x0 is referred to as the time component. No diﬀerence is made between upper or lower. in which case zi = ǫijk xj yk . we will use upper indices for the three-vectors. in this case an inertial system exists in which the two points are at the same space point and in that frame ds2 just represents the time diﬀerence ds2 = dt2 . the invariant length or distance (squared) is not positive deﬁnite. The Poincar´ transformations include e e Lorentz transformations and translations.. 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 . such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . The (invariant) lengths often have special names. xi are the three space components. So we could have used all upper indices in the above equations. Λµn n T ν1 .3). ˆ ˆ (1. it is convenient to distinguish lower indices from upper indices.. One has x2 = xµ xµ = t2 − x2 . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t.24) The quantity gµν ≡ nµ · nν is the metric tensor.22) (1.νn can be conν structed. .

(1. ′ ′ . not aﬀected by Lorentz transformations. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . (1. because one has (1. starting from ǫ0123 = 1.36) Exercises Exercise 1. (1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.32) . (1. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . The length squared of ∂ is the d’Alembertian operator.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. p).Introduction 6 called covariant. g0 = g1 = g2 = g3 = 1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. ∂2 . ∂1 .30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). p) = (E.27) It is easy to convince oneself of the nature of the indices in the above equation. Note that in this equation one has on left. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’.33) (1. . ∂3 ) = ∂ ∂ ∂ ∂ . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . It is an invariant tensor.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .∇ .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. where m is called the mass of the system. (1. for instance. 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. .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.28) ν ν Note that gµ with one upper and lower index. a∞ = 4πǫ0 2 /me e2 . 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 . = ∂ . but more important has the same eﬀect on lefthandside and righthandside. Each of these permutations leads to a minus sign. = −24.34) (1. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . ∂t (1. constructed via the metric tensor itself. deﬁned by ∂2 − ∇2 .

(b) The coordinates (a0 . 0. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. 0. a2 ] we can ﬁnd basis vectors n− . a+ . a+ . 1. 0. Compare it with the proton mass and use Eq. 0) ˆ n2 = (0. ]. a2 ] or [a− . 0.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . Also construct and calculate the Planck length Lpl . (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). aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. σ} is a permutation of {0. One does not need to know . a1 . 0. ˆ ˆ ˆ ˆ These are n1 = (0. Also for the coordinates [a− . a+ . which is the Compton wavelength for the Planck mass. ρ.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept).(A · B) C using the properties of the tensor ǫijk given in section 1. [Note: what is the MKS unit for V/T?] Exercise 1.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). How are a+ and a− related to a+ and a− . .13 to give its actual value in kg. 1) ˆ . n3 . 1. n+ . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . 1. This demonstrates how a careful use of units can save a lot of work. Exercise 1. but only appropriate combinations. a1 . 0) ˆ n0 = (1. 0. . g+− and g−+ .2 Prove the identity A × (B × C) = (A · C) B . a1 . 0) ˆ n3 = (0.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . re = e2 /4πǫ0 me c2 to the Compton wavelength. c. g−− . n1 . n2 . α and me c2 = 0. ˆ ˆ . .3. ǫ0 . a2 . (d) Use the value of the gravitational constant GN / c5 = 6. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . 3} the index α can only be equal to one of the indices in ǫµνρσ . 2.511 MeV. me . e. ν. n2 . Exercise 1. 1. by a simple few-line reasoning [For instance: If {µ. a3 ) are the expansion coeﬃcients using the basis vectors n0 . n1 .

3 are not covariant. 2M (2. ∂t p −→ pop = −i∇. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. p2 = pµ pµ = E 2 − p2 = M 2 . t) = exp(−i k 0 t + i k · r). one obtains (2.7) k 0 = ± k2 + M 2 = ±Ek .6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. E = p2 . ∂t2 (2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). (2. p).3) Equations 2.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . namely plane waves characterized by a wave number k. (2. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite.2) Acting on the wave function one ﬁnds for a free particle. (2. t) = − ψ(r. 8 .1 The Klein-Gordon equation In this chapter. ∂t 2M (2.2. t) = 0.4) where M is the particle mass. 2 + M 2 φ(r. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. But the replacement 2.Chapter 2 Relativistic wave equations 2. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.5) with (k 0 )2 = k2 + M 2 . i ∂ ∇2 ψ(r. t) = ∂2 − ∇2 + M 2 φ(r. Although it is straightforward to ﬁnd the solutions of this equation.1 and 2. φk (r. t).

As we will see later both problems are related and have to do with the existence of particles and antiparticles. rather than as a wave function. just as the dependence on time was in ordinary quantum mechanics. This continuity equation can be written down o covariantly using the components (ρ. for which we need the interpretation of φ itself as an operator. (2. This leads to the existence of negative densities. 2. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. k) + ei k·x φ(−Ek .e. The appropriate current corresponding to the KG equation (see Excercise 2. there are solutions with negative energy. however. (2π)3 2Ek (2.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. (2.10) ∂t which follows directly from the Schr¨dinger equation. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. ∂µ j µ = 0.2) is j µ = i φ∗ ∂ µ φ or (ρ.Relativistic wave equations 9 i. And. there are no longer fundamental objections to mix up space and time. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. 2.14) ˜ ˜ Introducing φ(Ek .9) They satisfy the continuity equation. (2. Then. j) = ↔ i φ∗ ∂0 φ.8) (2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. we want the most general solution.5). 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. k) ≡ a(k) and φ(−Ek . j) of the four-current j. −i φ∗ ∇ φ . ∂ρ = −∇ · j. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . ↔ ↔ (2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x).15) . is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. The KG equation. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends.3).12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. −k) . which is what Lorentz transformations do. having the KG equation as a space-time symmetric (classical) equation.11) Therefore. but rather the zero component of a four vector. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. 2M 2M (2. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. relativistically the density is not a scalar quantity.13) ˜ with (in principle complex) coeﬃcients φ(k).

This shows how parity transforms k-modes into −k modes. x) → (t.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). (2π)3 2Ek (2. (2.14 and 2. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). Performing some Lorentz transformation x → x′ = Λx. for which we consider space inversion. Since a · b = a · ˜ it is easy to see that ˜ b. (2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . Another symmetry is found by complex conjugating the KG equation. x ∂µ ∂ µ + M 2 φ(˜) = 0. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0.e.15 one has elimated k 0 and in both equations k · x = Ek t − k · x.23) . (2π)3 2Ek (2. (2π)3 2Ek (2. 2.16) The consequence of this is discussed in Exercise 2.18) (2. after restricting the energies to be positive).22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). which implies (xµ ) = (t.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). i. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.17) (2.3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. 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) . (2.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . −x) ≡ (˜µ ). 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).Relativistic wave equations 10 In Eqs 2. φP (x) ≡ φ(˜) (P for parity).6 We will explicitly discuss the example of a discrete symmetry. changing the sign of the spatial coordinates.

Note that a(k) and b(k) are independent of t. This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. t). t). 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ). Exercises Exercise 2. t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. (2π)3 2Ek where Ek = Exercise 2. QV ≡ satisﬁes ˙ QV = − ds · j. letting V → ∞.4 ˜ Write down the 3-dimensional Fourier transform φ(k. d3 k F (Ek . Q = 0. t) for the mode expansion in Eq. t) ≡ d3 x φ(x. Exercise 2. ↔ ↔ Exercise 2. that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). For the real ﬁeld one has aC (k) = a(k). k) = (2π)4 k2 + M 2 .3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. is conserved. Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. t) = ei Ek t (i∂0 + Ek ) φ(k. ˙ and thus for any normalized solution the full ‘charge’.2 Show that if φ and φ∗ are solutions of the KG equation.e. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k). 2.1 Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume.15. S V d3 x j 0 (x). |f ′ (xn )| .Relativistic wave equations 11 i. k).

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

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

Next. In the same way we can consider rotations around the x. 2 (3. is a 3-dimensional Lie-group. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. (3.6) Summarizing. They satisfy the commutation relations [Li . which reﬂects itself in the noncommutation of the generators. Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . x]] + [z. x]).4) R .7) = a0 0 −1 i 0 1 0 0 1 • [x. y] ∈ A. z]] + [y.2 π) (-n.5) Rotations in general do not commute. namely that there exists a bilinear product [. y ∈ A ⇒ [x.z N N or (Lk )ij = −i ǫijk . These generators form a threedimensional Lie algebra SO(3). These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). . [y. Locally this is a 3-dimensional Euclidean space and SU (2). and thus (σ · n)2 = 1. One way of viewing the parameter space.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). thus is as the surface i=0 of a sphere in 4 Euclidean dimensions.1: the parameter spaces of SO(3) (left) and SU (2) (right). Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. Lj ] = i ǫijk Lk . 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. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . Ly and Lz . With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. y]] = 0 (Jacobi identity). • [x. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3.and y-axes 0 0 0 0 0 0 0 −i . ˆ therefore. [x. z ) = lim N →∞ that are generated by i 0 . ] that satisﬁes • ∀ x. x] = 0 (thus [x. for rotations around z for instance. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . π) (n. (3. 0 (3.Groups and their representations 14 (n. [z. ˆ R(ϕ. y] = −[y. with real a’s and 3 (ai )2 = 1.

in particular that AB → RAB = RA RB ). it is suﬃcient that the generators commute with H. . 2 2 2 (3. They satisfy σi σj σk = i ǫijk . 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. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones.e. .9) The parameter-space of SU (2). now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). . . 2 (3.e. i. for ﬁxed n. H] = 0 for g ∈ G. . with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. thus. immediately sees that the Lie algebras are identical. For instance in a function space translations.12) One. also can be considered as a ﬁlled 3-sphere. (3. the above mapping corresponds to the trivial mapping of the Lie algebras. ˆ ˆ A(ϕ.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics.11) Thus σx /2. . (3. In the full parameter space. . The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. i.13) 0≤ϕ≤π The relation A(σ · a)A−1 = σ · RA a (valid for any vector a) can be used to establish the homomorphism (Check that it satisﬁes the requirements of a homomorphism. 2! . n! 2! n=0 ∞ (3. i. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. n) −→ R(ϕ.15) [g.10) σ ˆ · n. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . n) i ∂ϕ = ϕ=0 (3. n) ˆ −→ R(2π − ϕ.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. [g. The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. SU (2) ≃ SO(3).e. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). sometimes in more than one way (but this will be discussed later). n) π ≤ ϕ ≤ 2π. Suppose we have a system described by a Hamiltonian H. however. n) ≈ 1 + i ϕ J · n. thus. Near the identity. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. 3. (3.14) For a Lie group.

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

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

0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ .g. G = 10 (3. σ/2 in a two-dimensional (spin 1/2) case.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. We considered the generators and looked for representations in ﬁnite dimensional spaces.26) (3. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. e To derive some of the properties of the Lorentz group.27) (3. (3. From this property. (ii) |Λ00 | ≥ 1. Writing x as a column vector and the metric tensor in matrix form. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. det Λ = −1 is called improper). proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.3 The Lorentz group In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . In this section we consider the Poincar´ group. consisting of the Lorentz group and translations.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . e. it is convenient to use a vector notation for the points in Minkowski space.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. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . 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.Groups and their representations 18 3.25) (3.22) (Note that xT is a row vector). it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. (det Λ = +1 is called proper.

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

we continue using covariance arguments.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .g. Also this group can be divided into four parts. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. [M µν . Of the four parts only L↑ forms a group. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . In order to ﬁnd the commutation relations including the translation generators. 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.g.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. 3. (3. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. a).37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . We will just require Poincar´ invariance for the generators themselves. Summarizing. There are six generators. P+ etc. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . ǫ). This is a + normal subgroup and the factor group is the Vierer group.33) These four transformations form the Vierer group (group of Klein).36) (3. 3 30 with η = ω and K = M (e. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν .40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . K = M ). Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. We therefore ﬁnd (again) that there are six generators for the Lorentz group. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis).38) Explicitly. inﬁnitesimally approximated by (I + ω. i i0 i i0 1 ijk M jk 2ǫ (3. The extension to the Poincar´ group is e ↑ straightforward.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. (3.39) (e. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .

that commute e among themselves. writing the generator P = (H/c.45) Note that the commutation relations among the generators M µν in Eq. For instance. 3. [J i .44) (3. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . K j ] = i ǫijk K k . Explicitly. [K i . ν (3. (3. (3. the generators J 2 and J 3 in the case of the rotation group. one obtains 2 [P i . These form a group called the little group associated with that four vector.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. [P µ . [J i . = pµ |p.41) 2 At this point. check that U1 U2 corresponds with Λ1 Λ2 ). P j ] = i δ ij H/c2 . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . 3. 2 2 (3. a) = Λ−1 µ P ν . . They just state that M µν transforms as a tensor with two Lorentz indices. For instance. (3. . (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν .38 could have been obtained in the same way. . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . a)P µ U † (Λ. .46) [K i . [J i . P µ |p. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. P ρ ] = 0.48) . (3. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. P j ] = [P i . The eigenvalues of these will be the four-momentum pµ of the state. P ν ] [M µν . Taking any one of the states in an irreducible representation.47) (3. a) = Λµν P ν or U † (Λ. P ) in terms of the Hamiltonian and the three-momentum operators. the generators J± in the case of the rotation group. . ǫ) = 1 − ωαβ M αβ + i ǫα P α . a)P µ U (Λ. . We have here reinstated c. H] = 0. The generators of the translations are the (momentum) operators P µ . The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). . known from non-relativistic quantum mechanics are obtained. other states in that representation are obtained by the action of operators outside the maximal commuting set. P j ] = i ǫijk P k . the generators M αβ no longer exclusively act in Minkowski space. because one then sees that by letting c → ∞ the commutation relations of the Galilei group.43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . as part of the set.42) (to decide on where the inverse is. H] = i P i . = −i (g µρ P ν − g νρ P µ ) . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. Inﬁnitesimally. leads to U (Λ. K j ] = −i ǫijk J k /c2 .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. H] = [J i . [K i . J j ] = i ǫijk J k .

50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . (3. = = i exp − ǫµνρσ M µν pρ sσ |p.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. . the third case represents the vacuum. 2 (3. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ .54) This will enable us to pick a suitable commuting ’spin’ operator. Pν ] = [Wµ . . .55) commute with all generators and therefore are invariants under Poincar´ transformations.49) (3. . which can thus be chosen spacelike. M ǫijk . i ǫµνρσ W ρ P σ . p0 > 0. . (3. 0. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. Wν ] = −i (gµα Wν − gνα Wµ ) . Wα ] = [Wµ . . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. Massive particles: p2 = M 2 > 0.53) (3. while the others have no physical signiﬁcance. .56) . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). .51) The following properties follow from the fact that W µ is a four-vector by construction. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3.52) (3. 2 exp (−isµ W µ ) |p. for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the ﬁrst two cases correspond to physical states. . 0.

In that case the vectors ni (p) are just the space directions and W = (0.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . (1. M J) with components W i = (M/2)ǫijk M jk . 0.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). 1. (0. . 0. In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 .e. (0. W j (p)] = i M ǫijk W k (p). 0).e. show that W /M can be interpreted as the intrinsic spin. W i (p) = −W · ni (p)). 0).Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. s. M2 (3.. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 0. |p3 |). 0).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 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. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 3. ms . The commutation relations [W i (p).61) one sees that S i (p) i =− W2 . 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. p. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. 0. .58) i. M2 (3. 2 . Together with the four momentum states thus can be labeled as |M. 0. . Having made a covariant decomposition. p0 > 0. 0. (3.57) (i. 1. 1 2 (3. (3. i (3. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). 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 = 0.64) such that in an arbitrary frame where the four vectors p. n1 (p).59) (3. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p).65) . it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 1.

P µ .1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. The vacuum: pµ = 0. W 2 and λ is derived from the general commutation relations for [Wµ .72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. 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). where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). 3.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p).Groups and their representations The above does actually include the massive case. Note that angular momentum does not contribute to helicity as L · p = 0. W 2 (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 . a)|0 = |0 . λ . (3. however. (3. 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 .13. 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. which can take any integer or half-integer value. U (Λ. But if M = 0 one has a diﬀerent algebra (Eq. and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0). This state is in physical applications nondegenerate and is denoted by |0 . 3. Thus we have P 2 . |p| (3.71) which is called the helicity. W 1 (p)] [λ(p). Wν ].73) (3. Exercises Exercise 3. We don’t know of any physical relevance. |p. thus.68) (3. and gives [W 1 (p). A massless particle.67 has a vanishing right hand side). The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. 1 2 24 (3. W 2 (p)] = = = i M 2 λ(p). i W 2 (p).67) (3. implying continuous values for the spin (actually for W i = M S i ). −i W 1 (p). Thus 2 λ(p) = J ·p . W µ (p) = λ(p) P µ . W 2 (p)] [λ(p). is characterized by its momentum and the helicity.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). The algebra of W 1 . This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a).

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

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

With this choice of parameter-spaces the plus and minus signs are actually relevant. It is simply connected and forms the covering group of L↑ . becoming the deﬁning representations of SL(2.6) (4. The matrices in SL(2. Special cases are the following representations: Type I : (j.5) = 1 (J + i K). C) is the group of complex 2 × 2 matrices with determinant one. 2 (4. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . 2 1 (J − i K). j ′ ). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ .3) (4. similarly as the homeomorphism between SO(3) and SU (2).9) (4. K = −i . K = +i . B j ] = i ǫijk Ak . i ǫijk B k . (4. U (ϕ) H(η) (4.7) From the considerations above. The group SL(2. They precisely correspond to the two types of representations that we have seen before.4) (4. j) K = −i J (B = 0). C).8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. B j ] = [Ai .1) (4.1 The Lorentz group and SL(2. (j. This tells us how to ﬁnd the representations of the group.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). Aj ] = [B i . They will be labeled by two angular momenta corresponding to the A and B groups. 0) Type II : (0.10) . C) Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (hermitean) combinations A = B which satisfy the commutation relations [Ai . respectively. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. K = i J (A = 0). it also follows directly that the Lorentz group is homeomorphic with the group SL(2. 0. 2 2 σ σ J = . (4.Chapter 4 The Dirac equation 4. 2 2 (4.

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

a) and ˜ a = (a0 . but rather the operators W i (p) should be used. (scalar). + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . where a = (a0 . The i angular momentum operators J are represented by σ i /2. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. they cannot be connected by a unitary 2 2 transformation. 2 2 (4. Within the Lorentz group. but to L↑ . (pseudoscalar).21) In nature parity turns (often) out to be a good quantum number for elementary particle states. etc. For a particle at rest only angular momentum is important and we can choose ξ(0. A parity transformation changes aµ into aµ . C) are suitable for representing spin 1/2 particles. It has the same eﬀect as lowering the indices. The behavior is the same for classical quantities. e. ǫµνρσ will change sign. 4.g. (vector).1. i. We also have seen that the representations (0. Thus M ∗ and M are not equivalent within SL(2.23) u(p. 0 χm uR H(p) . 4. therefore. 0). The representing member of the class P− is the parity or space inversion operator Is . C). 2 ) and ( 2 . we have seen that in the rest frame W i (p)/M → J i . (axial vector). Taking M (Λ) = H(p). 1 ) and ( 1 . m) = χm . m) = χm uL H(p) 0 . the aij elements of a tensor will not change sign.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. = (4. From the deﬁnition of the representations (0. −→ M (Λ) (4. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). generators. −→ −p (vector).20) 4.e. that have the correct commutation relations. This provides the easiest way of seeing ˜ what is happening. the well-known two-component spinor for a spin 1/2 particle. 1 ) and ( 1 .13. as we have seen in section 4. i. the boost in Eq. 0) of SL(2. but by a transformation belonging ↑ ↑ to the class P− . (4. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ .17. For the spin 1/2 representations of the Poincar´ group including parity we. we obtain for the two components of u which we will refer to as chiral right and chiral left components. ) −→ ( . they can be connected. 0) are inequivalent.e. −→ −r (vector). must combine the repe resentations. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. −a). m) = η(0.The Dirac equation 29 where The matrix I2 does not belong to L↑ . Although these operators cannot be used to label the representations.

4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. The ﬁrst indeed is solved. (4. (4. (iγ µ ∂µ − M ) ψ(x) = 0. The general requirements for the γ matrices are thus easily obtained.33) deﬁne for a four vector a the contraction a = aµ γµ / . one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). . (4. γ ν } = 2 g µν .32) suppressing on the RHS the identity matrix in Dirac space. this can be rewritten 1 µ ν {γ . The explicit realization appearing in Eq. It is of a form that we also played with in Exercise 2. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). 0 H 2 (p) uR uR . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). = H −1 (p). This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. (4.26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . (4.28) which is a covariant (linear) ﬁrst order diﬀerential equation.7.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. We will discuss another explicit realization of this algebra in the next section. = (4. From this one obtains the Cliﬀord algebra for the Dirac matrices.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. p (4.24) (4. Applying (iγ µ ∂µ + M ) from the left gives (4. Since ∂µ ∂ν is symmetric.29) which is an explicit realization of the (momentum space) Dirac equation.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . γ } ∂µ ∂ν + M 2 ψ(x) = 0.27 is known as the Weyl representation. 2 (4. {γ µ . ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. which in general is a linear equation in pµ .

4. The density j 0 = ψγ 0 ψ = ψ † ψ (4. γ ν } = 2 g µν .1: The Dirac sea of negative energy states and antiparticles. Relying on the Pauli exclusion principle for spin 1/2 particles.g. This hole forms an antiparticle with the same mass.3 General representations of γ matrices and Dirac spinors {γ µ .35) is indeed positive and j 0 can serve as a probability density. The second problem encountered before.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. Dirac supposed that the negative energy states were already completely ﬁlled and the exclusion principle prevents any more particles to enter the sea of negative energy states. (4. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. . remains.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!).37) Using the explicit form of γ in the Weyl representation (see Eq. opposite charge). From the vacuum a particle can be removed. This can most easily be seen in the particle rest frame.27). one sees that there are two positive and two negative eigenvalues. Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. E = +M (twice) and E = −M (twice). 4. (4. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. The Dirac sea forms the vacuum. (4. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.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 . the one of the negative energy states.34) for the adjoint spinor ψ ≡ ψ † γ0 .36) (4. . This problem was solved by Dirac through the introduction of a negative energy sea. but with properties such that it can be annihilated by the particle (e.

42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.R. =ρ ⊗1= 0 −1 3 (4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . γ σ ].47) (4.R.R. (γ5 )S. −→ Sγµ S −1 . We note that ψ → Lψ and ψ → ψ L−1 .R. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. (γµ )S. is 1 1 1 (4. γ 2 ].41) (4.9 and 4. For example.40) (4. S −1 . γ 0 ]. For instance in the Weyl representation (but valid generally). 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 . 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . the rotation and boost generators in Weyl representation in Eqs 4. = S(γµ )W. γ µ } = 0 and explicitly one has 0 1 1 . (4. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .The Dirac equation 32 The Weyl (or chiral) representation: 0 1 .45) (γ5 )W. 2 (4. −→ Sψ. which in the case of the Weyl representation are just the upper and lower components. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . 0 (4. that project out chiral right/left states.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .3) .46) (4.48) are projection operators. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. 4! (4.43) S=√ .44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . while L−1 γ µ L = Λµ γ ν .49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .

−iσ 2 0 −ǫ = . is a left-handed spinor. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. T (4. (C)S.59) −→ ψ c = C ψ = η ǫ ξ∗ . in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . e. Explicitly.58) where ηc is an arbitrary (unobservable) phase. In any representation a matrix C exist. that brings ψ → ψ c .55) (4. but it does. x = (t. ∂ (4. which is again a solution of the Dirac equation. Starting with (i/ − M ) ψ(x) = 0 we note ∂ that by hermitean conjugating and transposing the Dirac equation one obtains iγ µT ∂µ + M ψ (x) = 0. The latter implies that the conjugate of a right-handed spinor.R.52) η ξ one can apply space inversion. parity.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. ∂ (4. for instance. ∂ with the solution ψ c (x) = ηc Cψ (x).R.R. Some properties of C are C −1 = C † and C T = −C.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. usually to be taken unity.g. x) → x = (t. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.53) (4. C is real and one has C −1 = C † = C T = −C c and [C.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. 0 −ǫ 0 0 −ǫ 0 . γ5 ] = 0. T (4. = = 0 iγ γ = −iρ ⊗ σ = −iσ 2 −iσ 2 iγ 2 γ 0 = −iρ3 ⊗ σ 2 = 0 2 0 1 2 Thus we ﬁnd back the Dirac equation.6).56) (i/ − M ) ψ c (x) = 0. (i/ − M ) ψ(x) = 0.57) (4. This symmetry does not change the spin 1/2 nature. reverse the charge of the particle.R. ψ= (4. ψR .54) e. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . called charge conjugation.g. = 0 ǫ iσ 2 (4. As with parity we look for a transformation. ψ= (4. such that T Cγµ C −1 = −γµ . (or W. One has ψ c = −ψ T C −1 . In S.) and all representations connected via a real (up to an overall phase) matrix S. (C)W.

4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. (2π)3 2Ek (4. ψ(x) = u±s (p) e−i p·x .62) where χs are two independent (s = ±) two-component spinors. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. The ¯ ¯ solutions are normalized to u(p. s) and C v T (p. s ) = 0. s) + v(k. s)¯(p. where u satisﬁes −σ · p Ep − M u(p) = 0. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4..65) (4. s)v(p.70) = Ps = ˆ 0 1∓σ·s 2 2 . s ) = v † (p. s′ ) = 2Ep δss′ . ⇔ (/ + M ) v(p) = 0. ψ(x) = v ±s (p) ei p·x . ∂ −iσ · ∇ −i ∂t − M (4. s ) = v (p.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions.61) σ·p −(Ep + M ) There are also two negative energy solutions. ¯ (4.The Dirac equation 34 4. s) e−i k·x b(k. the spin polarization vector in the rest frame is the starting ˆ point. v(p. s) ei k·x d∗ (k.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. the best representation to describe particles at rest is the standard representation. ¯ ¯ † ′ u (p. s′ ) = −2M δss′ . ⇔ (/ − M ) u(p) = 0. 2M 2M −/ + M p v(p. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. s)v(p.64) (4. s) u = . s′ ) = 2M δss′ . (4.68) (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. In an arbitrary frame one has s · p = 0 and s(p) can e.g. s) v = . s)¯(p.60) ψ(x) = 0.1). the spinors satisfy C uT (p. v (p. s) = v(p. s) = u(p. with E = Ep = + p2 + M 2 .63) u(p. 2M 2M (4. In that frame is a spacelike unit vector sµ = (0. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. p (4. s) . s) = Ep + M σ ·p Ep +M χs . s) = u(p. s). s)u(p. s)v(p. be obtained by a Lorentz transformation. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. s)u(p. ¯ ′ ′ Explicit solutions in the standard representation are χs . Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s). s)u(p.69) = − s In order to project out spin states. 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.

p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. u(p. λ) = u(p. 4. . etc. Note that for solutions of the massless Dirac equation /ψ = 0. p ψR/L ≡ PR/L ψ are independent solutions. helicity states. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . λ = −1/2) = √ − E+M √ 0 0 E+M (4. λ = − 1 ) = √ E + M − E − M . This means that in the solution space for massless fermions the chirality states. chirality. λ). +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 ). v(p. (4. v(p. one has in standard representation: 0 E+M E+M 1 1 0 . −p3 −p1 − i p2 1 v(p. The chirality projection operators in Eq.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . For instance with the z-axis as spin quantization axis. In principle massless fermions can be described by twocomponent spinors. λ = +1/2) = v(p. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . u(p. 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. Explicit examples of spinors are useful to illustrate spin eigenstates. λ) and C uT (p. Therefore. λ = + 1 ) = √ + E +M − E −M . (4. λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . while the negative helicity (λ = −1/2) is in essence lefthanded. λ = −1/2) = u(p. −1/2) = √ E + M −(E + M ) 0 . 2 2 0 √ 2 √ 2 − E+M − E−M 0 . By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components).48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). √ 0 u(p. γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0.74) . λ = − 1 ) = √ √ v(p. λ) = v(p. +1/2) = √ 0 E +M E+M .73) Also for helicity states C uT (p. We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded.71) p − i p2 . 1 u(p. −1/2) = √ (4.

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

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

you need to determine the corresponding conﬁnement condition for ψ. Note: use ﬁrst the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions. i. Note: To prove this. there is no current ﬂowing through the surface of the cavity. In this solution the wave number k = |k| √ = E 2 − M 2 and χm a two-component spinor while j0 and j1 are spherical Bessel functions. r). M R = 1.8 × 10−6 . Calculate for now just |M |2 . The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. Excercise 4. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. (b) Next. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.66 MeV.The Dirac equation 38 Exercise 4. The pion has spin 0.e. Do you understand why? The calculation of the life-time is completed in chapter 11. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e .511 MeV). is a stationary solution of the (free) Dirac equation. mµ = 105. We will not pursue this chiral freedom at this point and take α = 0. n It turns out that there is actually a whole family of conditions that provide conﬁnement. Calculate the value of k for M R = 0.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). .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 ). namely i / ψ = eiαγ5 ψ. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. ν p Note: how such parametrizations work will be explained in chapter 8. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. we want to conﬁne this solution to a spherical cavity with radius R. me = 0.57 MeV.5 and M → ∞. the decay would even be forbidden. For such a cavity nµ = (0. 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 ).

M2 (5. λ) + ǫ(λ)∗ (k) eik·x c∗ (k.2) (5.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors.1) The Lorentz condition implies that k µ ǫµ (k) = 0. In an arbitrary frame ǫµ (k) with λ = 0. just as the two-component Dirac spinors did for spin 1/2. The vectorﬁeld is therefore (λ) suited to describe spin 1.4) λ=0. A). These can describe spin 1 states in the rest-frame.2 The electromagnetic ﬁeld F µν = = ∂ µ Aν − ∂ ν Aµ 1 −E 2 0 −E 1 E 0 −B 3 2 E 3 B 0 3 E −B 2 B 1 Aµ = (φ. These are solutions of the Klein-Gordon equation with in addition a condition.1 Fields for spin 1 In section 3. namely ǫµ (k0 ) = (0. (2 + M 2 )Vµ = 0. ǫλ ). µ (2π)3 2E µ (5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. (5. λ) . In the restframe. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. there are then indeed three independent possibilities. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .5) −E B2 .7) . 39 (5. 0).± 5. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .6) where Aµ is the four-vector potential (5.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation.Chapter 5 Vector ﬁelds and Maxwell equations 5. with ∂µ V µ = 0 (Lorentz condition). ±(1) are obtained by boosting the restframe vectors.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. where k = k0 = (M.

2Aµ = 0.8) (5. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. in accordance with what we found in section 3. 5.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. ∂µ Aµ = 0. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. As remarked above. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). Under a gauge transformation Aµ −→ Aµ + ∂µ χ. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. One possible choice is the gauge A0 = 0 (5.15 the d’Alembertian is replaced by 2 + M 2 . moree over.16) (5. 0. 40 (5. in which case one has 2Aµ = jµ . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. (5.13) Hence. it will vanish at all times. (5. For electrodynamics one has furthermore the freedom of gauge transformations. for M = 0. (5. i.11) This equation is not aﬀected by a gauge transformation. 0).12) (5. j) the Maxwell equations read ∂µ F µν = j ν . |k|). transverse or Coulomb gauge). The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld.10) the electric and magnetic ﬁelds are unchanged. −i. Although the Lorenz condition is a constraint. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.e. 0.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). (5. AP (x) = Aµ (˜). . if ∂µ Aµ = 0 for large times |t|.17) (radiation. The equation in vacuum.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. Therefore if k µ = (|k|. ∓1. only two independent vectors remain ǫ(±) µ (k) = (0.15) (5. (5. It states the absence of magnetic charges and Faraday’s law. ˜ x Note that via parity transformation one obtains another solution. if in Eq.5. Taking the divergence of this equiation.14) of which the solutions give the Li´nard-Wiechert potentials. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . Furthermore.

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

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

. ∂t ∇ · B = 0.9 explicitly give the equations ∇ · E = ρ.Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5. ∂t ∂B = 0.8 and 5. 5.1 Show that Eqs. ∇×B =j+ ∇×E+ ∂E .

˙ 2 (6. ∂x (6. x) ˙ (6. d dt ∂L ∂x ˙ = ∂L .1) and as an example the lagrangian L(x.3) (6.5) with ∆x(t) = δx(t) + x(t) δτ . As an example. x) = K − V = ˙ 1 mx2 − V (x). thus t′ = t + δτ.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. These equations can be obtained from a lagrangian using the action principle.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum.Chapter 6 Classical lagrangian ﬁeld theory 6.4) 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 (6.6) The ﬁrst term leads to the Euler-Lagrange equations. and the total change x′ (t′ ) = x(t) + ∆x(t).8) 44 .1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). recall classical mechanics with the action t2 S= t1 dt L(x. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. ∂x ˙ (6. ˙ p= ∂L . (6. x (t) = x(t) + δx(t). Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current.

6. think for instance of a temperature or density distribution or of an electromagnetic ﬁeld).t2 ) d3 x . δ(∂µ φ) (6.9) which is done because the ﬁrst term (multiplying ∆x. R (6. We will come back to it in a bit more formal way in section 9. The hamiltonian H is deﬁned by H(p. + ∂L ∆x − H δτ ∂x ˙ t2 . x). dσµ . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ .e. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). t1 ) and R3 . . ∂µ φ. t1 (6. ∂µ φ(x)). (6. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . . = φ(x) + δφ(x) (6. i. Consider a lagrangian density L which depends on these functions. ∂µ φ(x)) = R d4 x L (φ(x).13) (6. their derivatives and possibly on the position. but which in 3 dimensions includes conserved quantities related to rotations and boosts. t2 ). ∂µ φ′ .6 can be rewritten as δS = .14) (6. ˙ (6.12) Here R indicates a space-time volume bounded by (R3 . (R3 . x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). L (φ(x). ˙ x The second term in Eq. . In general any continuous transformation. H is a conserved quantity. . ∂µ φ(x). In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds.2.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. .t1 ) (6. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. x) ≡ p x − L . λ=0 (6.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . 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 . .18) 1 Taking a functional derivative. which in classical mechanics vanishes at the boundary) does not play a role. .15) d4 x′ L (φ′ .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). indicated with δF [φ]/δφ should pose no problems. Variations in the action can come from the coordinates or the ﬁelds.11) λ=0 of which p and H are the simplest examples related to space and time translations. . . with ∆φ(x) = δφ(x) + (∂µ φ)δx . − d3 x . = ∂R (R3 . ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 .Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. x′ ) − d4 x L (φ. .

(2 + M )φ (x) = 0. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). 2 ∗ (6. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. 2 ψ = 0. One easily obtains (2 + M 2 )φ(x) = 0. (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. 2 (6.27) (6.25) (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).22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2.20) (6.29) . δL δL ∂µ = . ∂ (6. The integrand of the ﬁrst term must vanish.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. the second term is irrelevant. (6. (6.2 Lagrangians for spin 0. Using the variations in ψ (in the symmetric form).23) (6.19) δ(∂µ φ) δφ 6.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.24) (6. leading to the Euler-Lagrange equations.21) which diﬀer only by surface terms. leading to (2 + M 2 )φ(x) = 0. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities.

30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. that the lagrangian separates into two independent parts for M = 0.38) (6. The Majorana case is in fact more general. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . 2 (6. (6.36) (6. 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. (6.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.g.and lefthanded ones. the mass term in the Dirac lagrangian 6. We note that of which the left part coincides .g.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).31 is given by LM (Dirac) = −M ξ † η + η † ξ . Peshkin and Schroeder or Exercise 12. namely2 LM (Majorana) = + (6. 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 . . ∂ ∂ 2 2 (6.34) (6. ψL ≡ (ψL )c = C ψL = 0 (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . 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. The reasons for Grassmann variables will become clear in the next chapter. and thus (β ∗ α)∗ = α∗ β = −βα∗ . or equivalently separate the ﬁeld into right.31) showing e. 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 . (6. 2 αβ = −βα.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . but note the factor 1/2 as compared to the Dirac lagrangian. 1 M η † ǫη ∗ − M ∗ η T ǫη . In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).7).Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. Using the twospinors ξ and η.32) There exists another possibility to write down a mass term with only one kind of ﬁelds.

+ ∂R dσµ dσµ ∂R ≡ where d x . consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). (6. which is the space-integral over the zero-component of a conserved current. These do not aﬀect the dynamics and will give rise to conserved currents.47) .40) 6.46) 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 . Q= d3 x J 0 (x. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . the presence of a symmetry that leaves the lagrangian invariant. (6. t).3 Symmetries and conserved (Noether) currents Next. discussed in the next chapter.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .41) (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . Returning to δS the surface term is rewritten to δS = R d4 x . In particular when the lagrangian is invariant under symmetries. ∂µ J µ (x) = 0. As an example for the classical ﬁeld. (6. (6. we consider the second term in Eq. In this case.σ1 with σ = (R3 . Q(t1 ) = Q(t2 ). . these conserved quantities become operators. one can consider the variations at the initial or ﬁnal surface.18..44) δL ∆φ + Θµν (x)δxν . requires the presence of a conserved quantity.. δ(∂µ φ) For continuous transformations.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . 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. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. (6.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). 6. δ(∂µ φ) (6. which in a consistent picture precisely generate the symmetries.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . t). . δ(∂µ φ) (6.

Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . ↔ (6. 2 (6. t) is the conserved charge (Q = 0). The integral over the zero-component.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space).52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters. e. From Noether’s theorem one sees that the current Θµν ǫν is conserved. (6. = 0.53) These are the energy and momentum.g. .51) (6. H = d3 x Θ00 (x). 1 (6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. we obtain (since δxµ = 0).54) (6. Summarizing. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). (6. (6. ˙ Q = d3 x j 0 (x. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.50) (6.49) These currents are conserved as discussed already in chapter 2. In the next section we will see the quantity show up as the charge operator. 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.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν .3) j µ = i e φ∗ ∂µ φ. (6. 6. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). including e the space-time translations and the Lorentz transformations.

∂µ T µν (6. In general this tensor is not symmetric. however. ∂µ φ(x) + i e ∂µ Λ(x) e (6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .67) i.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4.70) . 6. • Vector ﬁeld Aµ : −i Sρσ = aρσ . the angle of the transformation depends on the space-time point x.66) In section 6. (6. where H ρµν is the quantity appearing in M ρµν .65) M µρσ = T µρ xσ − T µσ xρ . (6.g.e. is not invariant under local gauge transformations. (6.3 we have seen global transformations or gauge transformations of the ﬁrst kind. one has a tensor that satisﬁes (exercise 6.3). e. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. φ(x).64) (6. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. (6. Deﬁning T µν = Θµν − ∂ρ Gρµν . 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 .61) A ﬁnal note concernes the symmetry properties of Θµν . (6. −i eΛ(x) ∗ i eΛ(x) (6.68) (6.58) (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 .6) T µν = T νµ . 2 (6. 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. in which the U (1) transformation involves an angle e Λ.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.69) i eΛ(x) φ (x).63) µν = ∂µ Θ .5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). Any lagrangian containing derivatives.59) (6. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) 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). 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. 1 L (φ. As an example consider the Dirac lagrangian. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. −e j µ Aµ = −e ρφ + e j · A. Dµ φ) − Fµν F µν .76) We note ﬁrst of all that the coupling of the Dirac ﬁeld (electron) to the vector ﬁeld (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ .71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. (6. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x). Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector ﬁelds to the original lagrangian therefore produces a gauge invariant lagrangian. ∂µ φ) =⇒ L (φ. ∂ 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. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). 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 .72) (6. D ∂ A (6. 4 (6. (6.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance.73) (6.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.74) 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 .78) . The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ.

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

t). t) = Enr ψu (r. Comment: this relation is known as the Pauli equation.Classical lagrangian ﬁeld theory 53 Exercise 6. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. 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. This relation is known as the Pauli equation.5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 . 2me Exercise 6.6 (optional) Prove the properties of the tensor T µν in section 6. .4. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . in order to show that Θρσ − Θσρ = ∂µ H µρσ . A) (see also exercise 6. 2M 2M with Enr ≡ E − M ≪ M . use ∂µ T µν = 0 and ∂µ M µρσ = 0. Hint ≡ −gs µe · B = −gs Qe s · B.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 .

dq dp dq dp (7. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. 1 qx (t) = 3 d3 x φ(x. The bilinear product for the conserved quantities. ˙ L(q. ˙ 2 (7. labeled by the position. namely [q.2) H(p. (7.3) For variations δO(p. Further. H]P dt and dO = [O. q) = ˙ 1 mq 2 − V (q). is the Poisson bracket [A.5) ∆ x ∆3 x 54 .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.Chapter 7 Quantization of ﬁelds 7. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. q(q. (7. Q]P . q) = p q − L(q. 6. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. in our example p = mq. t). p]P . An immediate generalization for ﬁelds can be obtained by considering them as coordinates. there are conserved ˙ ˙ quantities Q (Eq. B]P = dB dA dA dB − . however.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t). Quantization of the theory is achieved by imposing canonical commutation relations between q and p. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. q) also considered in the previous chapter. p]P = 1. starting from the lagrangian L(q. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. p] = i.11) found through Noether’s theorem. dλ Examples are dO = [O. [q. (7.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q).

Π(y. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). Π(y. t). t)] = [Π(x. discussed in the section 6. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). t)] = i δ 3 (x − y).2. 2 2 2 (7. a) φi (x)U (Λ. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . t).6) ∆3 x L x (qx . ∂ qx ˙ ∂ qx ˙ (7. formally. (7.7) d3 x L (φ(x). t).11) (7.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. albeit with ’sharp’ functions.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). The essential conditions to consistently quantize a theory then are the following. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). i U † (Λ.Quantization of ﬁelds 55 etc. • Poincar´ invariance e The full variation of the ﬁelds. the discretization procedure above is an explicit example. Φ(f ) = d4 x φ(x) f (x). φ(y. a) = Rj (Λ−1 ) φj (Λ−1 x − a). ˙ δ φ(x) (7. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. As an example. qx+∆x ) . Sometimes it is useful to keep in mind that. qx . This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ.15) with furthermore the relations [φ(x. ∂µ φ(x)) = ˙ d3 x L (φ. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. ∆φ. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 .9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν .8) (7. ∇φ) (7. t)] = 0. for symmetry transformations.16) . φ. for the scalar ﬁeld theory.14) In fact. (7. we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ.12) (7. (7.

25) .19) 7.Quantization of ﬁelds 56 or if U (Λ.4. Mµν ] = i i∂µ φi (x). sometimes referred to as second quantization. P ] = 0 (7.23) The hamiltonian in this case can be expressed in the number operator N = a† a. ω a† a + 2 2 ω i − [Q. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. [φ(x). a† ] = a† . Pµ ] = [φ (x). (7.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . (7.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a.18) with S µν given in 6. a] = [a† . 2 i ωµν (S µν )i . [N. Q] = [P.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. Φ(g)] = 0. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν .17) (Sµν )i φj (x). 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. a† ] = 1. P ] = i. Microscopic causality implies local commutativity. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). and [N. P ] 2 (7. [Q. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . j (7.24) It is straightforward to ﬁnd the commutation relations between N and a and a† . φ(y)] = 0 if (x − y)2 < 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. a] = −a. [a. 2 2 (7. The measurements cannot inﬂuence each other or [Φ(f ). (7. a† ] = 0. This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the ﬁelds through Noether’s theorem. j 2 one has [φi (x).

In this way one has found for the harmonic oscillator the spectrum of eigenstates |n (with n a non-negative integer) with En = (n + 1/2)ω. (2π)3 2Ek (7. Note that we will often write a(k) or a† (k).Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n.32) (7. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. and (7. a(k ′ )] = [a† (k). The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . Π(x )]x0 =x′0 = 0. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2.g. φ(x )]x0 =x′0 = [Π(x). † √ a and a act as raising and lowering operators. a† (k ′ )] = 0. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . [φ(x). The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion.e. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .30) [φ(x). which represents the ’number of particles’ with momentum k.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . (n − 1) a|n † i.like’ commutation relations between a(k) and a† (k ′ ). Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.28) (7. 7.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). is equivalent with [a(k). a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). the real scalar ﬁeld φ(x) becomes d3 k (7. and we see that a state |0 must exist for which N |0 = a|0 = 0. e.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. (7.e.26) a(k) e−i k·x + a† (k) ei k·x . 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 .33) . i.

and multiparticle states |(k1 )n1 (k2 )n2 . (2π)3 2Ek (7. This term will be adressed below. (2π)3 2Ek (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 .37) (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.45) (7. . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). .40) This procedure is known as normal ordering. is solved by subtracting it as an (inﬁnite) constant.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).36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). 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). writing all annihilation operators to the right of the creation operators. . i. d3 k a(k) e−i k·x . |0 . n1 ! n2 ! (7.41) (7. (2π)3 2Ek (7. |k = a† (k)|0 . which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . assuring that the vacuum (by deﬁnition) has eigenvalue 0!.46) (7. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). a(k)|0 = 0. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .e.39) The problem with the vacuum or zero-point energy. 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. k 0 = Ek = k2 + M 2 ).38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. deﬁning particle states |k = |k (with positive energy. which now contains an inﬁnite number of oscillators.34) where V = (2π)3 δ 3 (0) is the space-volume.42) (7.35) where the vacuum contribution disappears because of rotational symmetry. (2π)3 (7. (2π)3 2Ek (7.44) p|φ− (x)|0 = ei p·x . .

(7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .50) or as integrals over d4 k. (2π)4 (7.52) The result for the invariant commutator function is [φ(x). x) = 0 and hence ∆(x) = 0 for x2 < 0.56) = −δ 3 (x). In order to calculate [φ(x).18). The last item to be checked for the scalar ﬁeld are the causality condition. (iii) ∆(0.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). φ(y)] consider µν [φ+ (x). φ− (y)] ≡ i∆+ (x − y).51) (7. d3 k ei k·(x−y) = −i∆+ (y − x). a† (k ′ )] 3 2E ′ (2π) k [φ− (x). 2π (7. ∆(x) = − ǫ(x0 ) δ(x2 ). i.e. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. (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. that they satisfy the required commutation relations in Eqs (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.48) (7.53) t=0 .49) (7. (7.17) and (7. = − (2π)3 2Ek (7. φ+ (y)] d3 k e−i k·(x−y) .54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . d3 k = (2π)3 2Ek = (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. 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 ). i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (2π)3 2Ek ≡ i∆− (x − y).

64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). ↔ (7. = ∂{µ φ∗ ∂ν} φ − L gµν . (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .63) i. 2 From the lagrangian density (7.57) (7. (7. which is equivalent to the relations (only nonzero ones) [a(k). The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x .65) 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) .61) (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x).66) .4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). (2π)3 2Ek (7.e. φ† (y)] = i∆(x − y). (2π)3 2Ek (7. extending it with the charge operator and the introduction of particle and antiparticle operators.62) (7. Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) . we will consider it as a repetition of the quantization procedure. a† (k ′ )] = [b(k). particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.59) (7.60) = i φ∗ ∂µ φ.Quantization of ﬁelds 60 7. (2π)3 2Ek (7.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).

s).68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). s)ei k·x . {b(k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (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′ )} = {d(k. jµ Θµν = = ψγµ ψ. d† (k ′ .76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. (2π)3 2Ek (7.67) i ψ / ψ − M ψψ gµν . s)v(k. The quantized ﬁelds are written ψ(x) = s d3 k b(k.e. interchanging creation and annihilation operators. † {ψi (x). . s)u(k. The solution is the introduction of anticommutation relations. s) e−i k·(x−y) j s † + vi (k.77) Note that achieving normal ordering. u v (2π)3 2Ek (7. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. i.70) where the last line is obtained by using the Dirac equation.71) (7. b† (k ′ . (7. s)vj (k.74) (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . ∂ 2 ↔ (7. s)e−i k·x . ↔ i ψγµ ∂ν ψ − 2 (7. s)ei k·x + d(k. s) ei k·(x−y) x0 =y 0 .78) (7. s)e−i k·x + d† (k. s). (2π)3 2Ek d3 k b† (k.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. s)¯(k.69) (7. s)¯(k. s)u† (k.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : .73) (7.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. (7. (2π)3 2Ek (7.75) (7.

ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y). but has the problem of being noncovariant. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . −λ(∂ρ Aρ ). (7.83) From the lagrangian density the canonical momenta are = = (7.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. ˙ δ Ai (7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. 4 Π0 Πi δL = 0. ∂ (7. For the scalar combination {ψi (x). 7.85) which reﬂects the gauge freedom. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. ψ j (y)] = (i/x + M )ij i∆1 (x − y). as the vanishing of Π0 induces a constraint. ∂ (7. 4 2 This gives the equations of motion discussed before. (7. ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem).87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.6 The electromagnetic ﬁeld 1 L = − Fµν F µν . one has † {ψi (x). we would have obtained [ψi (x).82) where i∆1 = i∆+ −i∆− . ˙ δ A0 δL = F i0 = E i .88) .84) (7.86) (7.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. ψ j (y)} at arbitrary times one has {ψi (x).80) 3 = δ 3 (x − y) δij .

pj ]P . j = 1. and are equivalent with [c(k. 0) .89) with four independent vectors ǫµ . Q]P . 7. containing a time-like photon. rj ]P . where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. 0)c(k. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. Πν (y)] = 0. what are [ri .e. These problems are solved by the Lorentz constraint between physical states given above. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.3) satisfy the conditions of a Lie algebra. ′ (7. 0. (7. Aν (y)]x0 =y0 = i gµν δ 3 (x − y).e. q) that dO/dλ = [O.91) (7. λ)|A = 0. a longitudinal photon and two transverse photons. i. [ℓi . ˙ ˙ . In fact the commutation relations imply ˙ [Aµ (x). 2.94) Choosing k µ = (|k 3 |.95) Exercises Exercise 7. λ). Aν (y)] = [Πµ (x).11. λ)e−i k·x + c† (k. for which it is suﬃcient that ∂µ Aµ |A = 0. It gives 3 k µ ǫ(λ) (k)c(k. pj ]P . The canonical equal time commutation relations are [Aµ (x). Πν (y)]x0 =y0 = i gµν δ 3 (x − y). µ λ=0 (7. and [ℓi . λ)c(k. (7. λ)ei k·x . which reproduce the Hamilton equations. 6. 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. (7. (d) Use Poisson brackets to calculate q and p. i. where Q is the conserved quantity in Eq.Quantization of ﬁelds 63 for physical states |A and |B .1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). 0. (b) Proof for a quantity O(p. µ λ=0 (7. c† (k ′ .93) which does exhibit problems with the time-like photon.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. λ) − c† (k.

Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. (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. (e) Show that ∆R (x) = 0 if x0 < 0.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. f (Λk) = f (k) for Lorentz transformations Λ. (b) Check the above commutation relation [φ(x). ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration).Quantization of ﬁelds 64 Exercise 7. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). It is invariant. (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. (2 + M 2 ) ∆hom (x) = 0. (2 + M 2 ) ∆inhom (x) = −δ 4 (x).2 Prove that the equal time commutation relations between quantum ﬁelds φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ ) Exercise 7.4 (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. Pµ ] = i∂µ φ(x). e−i a requires [φ(x). ∆hom (Λx) = ∆hom (x) if f is an invariant function. . Give also the expression for ∆C . we can write for ∆R an integral where the k 0 integration remains along the real axis.

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

Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. r) −→ xµ ≡ xµ = (t.1: The behavior of classical quantities under P. writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). ˜ (8. 8. (iγ µ ∂µ − M ) ψ(x) = 0.1 Parity xµ = (t.1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x). The consequences of P. and C quantity t r xµ E p pµ L s ˆ λ = s·p P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ T -t r −˜µ x E -p pµ ˜ -L -s λ C t r xµ E p pµ L s λ 66 . time reversal (T) and charge conjugation (C). −r). We can determine A. T and C for classical quantities is shown in the table 1. Both ψ p and ψ satisfy the Dirac equation.2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. Table 8. x (8. parity (P). starting with the Dirac equation for ψ(x). T.

m).6) (check this for the standard representation. iγ µ† ∂µ − M ψ(˜) = 0.4) (8. −λ)u(k.9) (8. if helicity λ is used instead of the z-component of the spin m. m) = = ˜ u(k. λ)γ0 v(k.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). −λ).13) ψ(x) −→ ψ c (x) = ηC C ψ (x). one can also consider the scalar ﬁeld and vector ﬁelds. m). The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ) e−ik·˜ + d† (k. ˜ −v(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.11) (8. λ)u(k.5) (8. 8. m) γ0 v(k. the above operation reverses the sign of λ). x The latter behavior of the vector ﬁeld will be discussed further below. λ) Pop = ηP b(k. −λ) e−ik·x − d† (k. λ)γ0 u(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. λ) e−ik·x − d† (k. λ)u(k. λ) eik·x . −λ). −λ) e−ik·x − d† (k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. −λ)v(k.3) (8. γ x (8.e.10) i. ˜ Pop d(k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. (2π)3 2Ek (8. x (8. λ)v(k.7) = λ d k x x ηP b(k. λ) P −1 = −η ∗ d(k. . λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. In the same way as the Fermion ﬁeld. op P (8. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). (i˜ µ ∂µ − M ) ψ(˜) = 0. λ)v(k.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜).

λ) Cop = ηC d(k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. . (8. m).Discrete symmetries 68 the latter being also a solution of the Dirac equation.19) Cop d(k. we start with the complex conjugated x Dirac equation for ψ.16) d k ηC d(k.15) (where one must be aware of the choice of spinors made in the expansion. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). u(k.21) where A is a 4 × 4 matrix acting in the spinor space. i. Norm conservation requires Top to be anti-unitary2 . (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. λ) −1 Cop = ∗ ηC b(k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. The (time-reversed) Dirac equation becomes. λ). λ) eik·x . x (8. Top |φ = φt | = (|φt )∗ . it is antilinear1 . λ). The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. x (8. m) ¯ = = v(k. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.3 Time reversal xµ = (t.18) (8. λ)C v T (k. x x (8. 8.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). m). (2π)3 2Ek (8. (8. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . As time reversal will transform ’bra’ into ’ket’.17) 3 = λ This shows that −1 Cop b(k. r). λ)u(k. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation.14) (8. λ)C uT (k.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .e. m) ¯ C v T (k. λ) e−ik·x + b† (k. antilinear operator is anti-unitary if A† = A−1 . r) −→ −˜µ ≡ −xµ = (−t. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . x iγ µT ∂µ − M ψ(−˜) = 0. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. The same relations hold for helicity states. λ) e−ik·x + b† (k. as discussed in section 4). λ)v(k.

25) (check this for the standard representation). λ) e−ik·x (2π)3 2Ek (8. −p. ∗ ˜ v (k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ)v ∗ (k. λ).26) d k x x ηT b(k. λ)u∗ (k. λ| a. λ a P |a. λ)u∗ (k. p. −λ a T a.34) . λ). a state |a. λ| ¯ C |¯.28) (8.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ)u∗ (k. −p. λ)v ∗ (k. −p.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k.24) (8. λ) i γ5 C v(k. (8.33) (8. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. The result is = = λ (8. λ).31) (8. ˜ Top d(k. λ) eik·x + d† (k.29) In table 2 the behavior of particle states under the various transformations has been summarized. −p. x (8.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. there are 16 independent of these bilinear combinations. ˜ = 0|b(k)|A 8. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. −λ |¯. λ). p. p. op T (8. λ)iγ5 Cv(k.30) (8. λ) Top = ηT b(k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k.Discrete symmetries 69 Table 8. λ) T −1 = η ∗ d(k.32) (8. λ a |a. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ) eik·x + d† (k. λ) eik·x + d† (k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. p. λ) eik·˜ + d† (k. λ)iγ5 Cu(k. Since there are 16 independent 4 × 4 matrices. λ)v ∗ (k. λ |¯.

p′ . ¯ ′ ′ (8. The 16 combinations of Dirac matrices appearing above are linearly independent. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . γ5 γ µ or γ5 ).38) (8.40) where the coeﬃcients are functions of invariants constructed out of the momenta. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). F. 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). for the vector current V µ (x). s′ |V µ (x)|p. (q 2 ). p′ |V µ (x)|p = u(p′ )Γµ (p′ . ¯ In general the expectation value between momentum states can be more complicated. 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 .. of which the expectation value between momentum eigenstates can be simply found. (q 2 ). Note that the lagrangian density L (x) → L (−x) under Θ.Discrete symmetries 70 Table 8.36) (8. and T. σ µν . as well as under the combined operation Θ = PCT (see Table 3).g.37) (8. C. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :.39) where Γµ (p′ .41) 2M 2M . F. 2M (8. called form factors. γ ν ]. As an example consider the vector current for a point fermion. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).35) The x-dependence can be accounted for straightforwardly using translation invariance. 8. in this case only q 2 = (p′ − p)2 . In many applications the expectation values of currents are needed. (8. p)u(p) e−i (p−p )·x . C. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. (8. s . 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 ). p′ |V µ (0)|p = u(p′ )γ µ u(p). p) can be built from any combination of Dirac matrices (1. e. For instance for nucleons one has Γµ (p′ .. γ µ .5 Form factors Currents play an important role in ﬁeld theory.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P.

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

3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). p′ = (E. 4M 2 = F1 + F2 . p = (E. GE GM = F1 − Q2 F2 . Exercise 8. σµν q ν F3 (q 2 ) 2M . 0. (c) Show that if a term of this form would exist. |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 . 0. 0.2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . 0. (b) Show that in the interaction with the electromagnetic ﬁeld. Exercise 8. |q|). −|q|/2). 0.Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . (c) Show that hermiticity of the current requires that F1 en F2 are real. (b) Show that such a term is also not allowed by time-reversal symmetry. 0.

QH (t)|q. t′ ≡ q ′ |q ′ . t0 ) = H U (t. t ′ QH (t′ )|q ′ .4) ∂ U (t. |ψH (t) = |ψH . t . (9. t ′ |q . t0 ).Chapter 9 Path integrals and quantum mechanics 9. t′ . in which the states are time-independent.1 Time evolution as path integral U (t. (9. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. |ψS (t) = U (t.6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .2) (ii) Heisenberg picture. (9. t0 )|ψS (t0 ) . [AH . t ≡ q|q.3) (9. and the operators are time-dependent.5) (9. H]. AH (t) = U −1 (t. (9. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. Consider the two (time-independent) Heisenberg states: |q. t0 ) ∂t (9. t0 ) AH (t0 ) U (t. AS (t) = AS and the states o are time dependent.7) 73 . in which the operators are time-independent. ≡ 0.

t = q ′ |e−i H(t −t) |q .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 ′ . QS |q ′ ≡ q ′ |q ′ . [A. . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj .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 .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. tj+1 |qj . Then qj+1 . t′ |q.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q .10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . eA eB = eC with C = A+B+ (9. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . 2 12 12 (9. B] + [A. 2π (9. H(P. .8) Choose t as the starting point with |q = |q. (9.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . B]] + [[A. B].qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . tj+1 |qj . These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|.. tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj . qj )]) . ′ (9. 2π (9. (9.16) this. .11) (9. 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 . The purpose of this is to calculate the evolution for an inﬁnitesimal time interval.12) where the transformation between coordinate and momentum space involves q|p = ei p. t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . . n |q (9. t′ |q.14) . use the Campbell-Baker-Hausdorﬀ formula.. Q) expressed in terms of the operators P and Q.q . qj ). The hamiltonian is an operator H = H(P. which is a hamiltonian function in which the operators are replaced by real-numbered variables. B] + . Combining the two terms gives qj+1 . |q1 q1 |e−i H∆τ |q . ˙ 2π 1 1 1 [A. with the correction1 being of order (∆τ )2 .

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

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. with Dα F [α] = n dαn N F (αn ) (9.21) again a multidimensional integral. α(x) = n αn fn and K(x.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions.24) (9. (9. . the following integrals can be derived for a real symmetric or complex hermitean kernel K.23) Having deﬁned the Gaussian integral. y) = ∗ m. Consider the gaussian functional as an example.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . (9. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.25) (see Exercises).27) (9. Note that procedure (i) is an example of the more general procedure under (ii).22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . Dα F [α] = Dα F [α + β]. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. N (9. A useful property of functional integration is the translation invariance. 2 (9. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. 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.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.

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. δα(x) or in discretized form δ 1 ∂ ∂ with . however. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .g. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . dx dy θ∗ (x) K(x. (9. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. αn = δmn . y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . α(y) = δ(x − y).31) δ δ α = dy α(y) = dy δ(x − y) = 1. i. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1.37) 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. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. F [θ.29) (9. y) θ(y) . αy = δxy . θη = −ηθ.35) For applications to fermion ﬁelds. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . products of the same pair appear.34) (9. (9.38) . the Gaussian-type functionals. e. δα(x) δα(x) δ αKβ = dy K(x. n θn fn (x) dx θ∗ (x) θ(x) . In principle the deﬁnitions of functionals is the same.30) (9. (9. In the expansion of the exponential (from second to third line).32) (9. Examples of functional diﬀerentiation are (9.e. θ2 = 0. which vanish. δα(x) (9.33) (9.

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

(9. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. Q(tn )|0 . .67) = = φ0 (q.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) .69) The factor that has been divided out in the ﬁrst line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ). T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. J=0 (9.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9.6 Euclidean formulation The above expression (Eq. J (9. Better is the use ˜ of imaginary time t = −it. 9. t 0 0out |0in J . .71) 9. T ′ |Q. such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. The importance of Z[J] is that the time ordered product of operators can be obtained from it. t′ |q. dq dτ + J(τ )q(τ ) . t) (9. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .68) (9. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . Since we have (neglecting multiplicative factors).70) for the generating functional is in fact ill-deﬁned. . . thus T →i∞ lim q. T = φ0 (q. t′ ) q ′ . ˜ t′ → ∞. = (i)n 0|T Q(t1 ) . . T → i ∞ and T ′ → −i ∞. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to ﬁnal(’out’) situation in the presence of a source. δJ(tn ) hence the name generating functional. t). t|Q.

Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. + V (q). The diﬀerentiations in Z[J] and ZE [J] are related. det K 1 . 1 = √ . δJ(tn ) J = 0 ˜ t = it (9.2 Check the examples of functional derivation for real-. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . τ dq ˜ dt ≡ −L q. ZE [J] = = where LE q. . .75) = − LE q.73) d˜ LE q. (9.76) which is a positive deﬁnite quantity. δ n Z[J] 1 Z[J] δJ(t1 ) . i dq .g. where Kp are the eigenvalues of the matrix Kmn in K(x.n fm (x) Kmn fn (y). ˜ dt (9.74) − V (q) 2 (9. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. e.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. . for Grassmann valued functions. (9. complex. . ˜ ˜ ZE [J] δJ(t1 ) .and Grassmann-valued functions. det K Exercise 9. ensuring convergence for the functional integral ZE [J].72) (9. y) = ∗ m. Exercises Exercise 9. δJ(tn ) = (i)n J=0 1 δ n ZE [J] .1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. when L q. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). .

det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .78) (9.3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . where dy K(x. 2 (9.79) (9. . y) K −1 (y.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . z) = δ(x − z).Path integrals and quantum mechanics 84 Exercise 9.

.3) The Euclidean formulation is as before. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . ∂µ φ) − Jφ] (10. . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ.Chapter 10 Feynman diagrams for scattering amplitudes 10. x′ . t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . . x. . Π) (10. t′ |φ. ∂µ φ) + Jφ] (10. φ(x.4) J=0 . (10. xn ). i. . δJ(xn ) 85 ≡ G(n) (x1 .2) = ∞ −∞ d4 xE [LE (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 ) . .e. .1) Dφ exp i σ d4 x L (x) and T′ Z[J] = lim T → i∞ T ′ → −i∞ Dφ exp − Dφ exp i T d4 x [L (φ.1 Generating functionals for free scalar ﬁelds The generating functional for quantum ﬁelds is a generalization of the results in the previous section to a system with more degrees of freedom.

∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. . (10. . . .9) i.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). .3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. . (2π)4 (k 0 + iǫ)2 − E 2 (10. J(xn ).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).e. . .5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . . . xn ) = x2 xn .10) J=0 In order to determine ∆F . These Green functions appear in the expansion Z[J] = n in n! d4 x1 . Furthermore.8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .13) . (10. As discussed in the previous section (exercise 9. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. (10. xn ) J(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. .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . with Z0 [0] = 1. d4 xn G(n) (x1 . 2 2 → (10. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).

the advanced Green’s function ∆A (x) = satisfying ∆A (x. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).e. 10.22) (10. i. ∆(x) = − d4 k e−i k·x . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation.16) C where C is a closed contour in the complex k 0 -plane.14) They are solutions of the inhomogeneous equation. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. t) = 0 for t > 0.20) = exp − 2 where or (see ﬁg.15) e−i k·x d4 k . 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).18) Thus we see various solutions. ∆F = ∆C .1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (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. φ+ (0)] = (2π)3 2E are also shown in exercise (7. In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. the causal Green’s function ∆C (x) = and ﬁnally the Green’s function d4 k e−i k·x = 4 (k 0 )2 − E 2 + iǫ (2π) ∆(x) = PV d4 k e−i k·x . The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). . µ 1 1 • Firstly.4).23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . t) = 0 for t < 0. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C .19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . (2π)4 (k 0 − iǫ)2 − E 2 (10. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .17) d4 k e−i k·x . 4 k 2 − M 2 + iǫ (2π) (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. (10. (10.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). (2π)4 k 2 − M 2 (10.

x3 . we use the fact that i∆F (x − y) = G(2) (x − y) = 0|T φ(x)φ(y)|0 to calculate ∆F via the time-ordered product. We ﬁnd (see also section 7. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. the same result as above.3) i ∆F (x − y) = 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0 = i ∆C (x − y).24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function.25) x1 x2 x3 = x4 + + . . or diagrammatically (4) (10. we have G0 (x1 . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1.1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. . because this is the only quantity entering Z0 [J].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. = θ(x0 − y 0 ) 0|[φ+ (x). x2 .

x2 ) = i ∆F (x1 − x2 ) = G(4.g ) (x1 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. ≡ −i g d4 z (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 . L (φ) = L0 (φ) + LI (φ).33) (4) i −i d4 z LI g 4! d4 z 1 δ i δJ(z) 1 exp − i 2 J ∆F J 2 −3 ∆2 (0) + 6i ∆F (0) F + d4 x ∆F (z − x) J(x) 4 d4 x ∆F (z − x) J(x) 1 exp − i 2 J ∆F J .26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10.30) and in zeroth order G(0. x2 . (10. (10.29) This expression will be the one from which Feynman rules will be derived. x2 . (2) (0) (10.2 Generating functionals for interacting scalar ﬁelds d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10. x2 ) = G0 (x1 .g ) (x1 .27) When interactions are present.e. the generating functional can be written.31) G(2. i. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . (10. x3 .g 0 ) = G0 = 1.32) (10. x3 . Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar.Feynman diagrams for scattering amplitudes 89 10. x4 ) = G0 (x1 . Z[J] = = Dφ exp i exp i d4 z LI (10. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].34) Introducing a vertex point to which four propagators are connected.35) z .

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

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

thus β. . x1 G(4. (10. 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 . in = |α.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. implying the absence of tadpoles. out = δαβ ↔ SS † = 1. ∂µ ∂ µ + M 2 φin (x) = 0. in = p. For the free scalar theory. one ﬁnds for the connected 4-point Green’s function at order g. in|S = β.g. in into ﬁnal state particle states (out-states) |β. out = p|p′ . (2) The one-particle state is invariant (conservation of energy and momentum. (2) (10. .3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. out|p′ . x4 ) = c 1 x3 z (10. out = |p′ . .45) x2 as the only surviving diagram (see exercises). in ⇔ β. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . out| = eiϕ0 0. p. out .43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). in|S = 0. (10. . out|α. e. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . p′ . x3 . x4 10. translation invariance). x2 . out (suppressing except momenta all other 1 m quantum numbers). (10.Feynman diagrams for scattering amplitudes 92 i. out| ⇔ |α. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). in = p. in|S|p′ . out| and S † |α. .e. out|α. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). Proof: 0. out|p′ . in = S|α.47) . . in| (choose ϕ0 = 0). in|SS † |α. Next. Proof: α. in|p′ .44) J=0 For the interacting theory. pn . in = |p1 . in|S = α. the vacuum expectation value of the ﬁeld φ(x) is zero. this will be translated to the action on ﬁelds. φout and the interpolating ﬁeld φ(x). out . in = β. Sβα ≡ β. considered explicitly. in = p. then (n) (n) Gsc = Gc for n ≤ 3.g ) (x1 .

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. 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) (10.51) (10. (10. as it stands. z) φ(z). 2 ∂µ ∂ µ + M0 φ(x) = J(x). a causality condition that can be proven to imply the absence of interactions.e. (10.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. in|φin S → φin S = S φout (x). S is invariant under e Poincar´ transformations: U (Λ. (10. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . i. Proof: β. out|φout = β. φ(y)] = Z [φin (x). (2π)3 2E (10. which in a later stage (renormalization) will become more important. the source term Jφ in the lagrangian density is treated in the interaction picture. The relation between S and Z[J] To establish this relation. β. The relation between S-matrix and in. z) φ(z) d4 y d4 z ∆A (x − y) K(y. implies the strong (operator) convergence φ(x) → Zφin (x). The convergence therefore must be weakened to t→−∞ √ (10. φin (y)] = iZ ∆(x − y). 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. this can actually not be used as it would imply [φ(x). e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S.g.and out-ﬁelds is: φin (x) = S φout (x)S −1 . a)SU −1 (Λ. e.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) .54) .53) α|φf (t)β −→ Z α|φf (t)β in for normalizable states |α and |β and φf (t) ≡ d3 (x) f ∗ (x) i ∂ 0 φ(x) with f a normalizable solution of the Klein-Gordon equation (wave packet). The time evolution operator (see Eq.Feynman diagrams for scattering amplitudes 93 while φ satisﬁes the inhomogeneous Klein-Gordon equation with the bare mass M0 (this is the mass appearing in the lagrangian L0 ). in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. a) = S.52) where Z is a constant. 9. √ Although the above.

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.61) e φ+ f where the normal ordered expression : e φ f : is used. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. express it in terms of advanced and retarded Green’s functions. 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. eB ] = [A. z) d4 y d4 z ∆(x − y)K(y. φ(x). e e ] = [A. y) δ δJ(y) : F [J].55) δU [J] δJ(x) = t→∞ t→−∞ −→ −→ √ i φ(x) U [J] → i Z φout (x) U [J] √ i U [J] φ(x) → i Z U [J] φin (x). S U [J]] = √ Z δ S U [J].57) i [φin (x). we can.63) . B] (for the case that [A. one has [A. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. provided [A. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x).60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). φ(x). z) . : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. C] are c-numbers. δJ(z) (10. e = i. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y.58) In order to ﬁnd a solution to this equation note that. (10.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. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.59) (10. z) δU [J] . B + C]e e . B] and [A. (10.e. φ− f (10.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. B] is a c-number). just as we did for φ(x). φ(y)] f (y) e e . B]eB .62) where F [J] is some (arbitrary) functional. which is equal to the expression e in which the creation operators are placed left of annihilation operators. [A. t i Ut∞ [J] φ(x)U−∞ [J] (10.

. iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function. It is straightforward to check that the S-matrix element now precisely is given by the amputated Green’s function multiplied with the momentum space wave functions of the particles in initial and ﬁnal state (by which we refer to the quantities multiplying the plane wave e±i p·x in the ﬁeld expansion. which is obtained considering only connected diagrams.2). . ←− − → × G(n) (x′ . .+ 1 √ Z ∗ d4 x′ . . . u(p) and v (p) for fermions and ǫµ (p) for vector ﬁelds). . .) iKx . .64) Therefore. . . p′ . (λ) v(p). fp′ (x′ ) . . .Feynman diagrams for scattering amplitudes 95 while for J = 0 one has U [0] = 1.68) 10. an S-matrix element between momentum eigenstates in initial and ﬁnal states is found by considering those source-connected Green’s functions (action of δ/δJ on Z[J]/Z[0]) where the external sources J(xi ) are replaced by the particle wave functions (which are the result of acting with φin (xi ) on momentum eigenstates). They start with the propagator (i∆F (k)) in momentum space. using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . i. d4 x . . = . .e. . . xn ). . c (n) (10. see exercise 10. Usually we are interested in the part of the S-matrix describing the scattering. pn ).. u(p). . fp (x) . . ..67) where ∆(p) is the Fourier transform of the (full) propagator ∆F (x). . which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. Explicitly.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. y).e. . x. . (10. ∆(pj ) j=1 (10. . Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . . . S =: exp 1 √ Z d4 x d4 y φin (x) K(x. . |S|p.δ. Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. k = k2 i − M 2 + iǫ . . (2π)3 2E n (10. iKx′ . . Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. . J=0 (10. . n (2π)4 δ 4 (p1 + .. . . . 1 for scalar case. . . + pn ) G(n) (p1 . pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . .65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. ¯ ¯ and the amputated or one-particle irreducible (1PI) Green’s functions n −i (n) Γ(n) (p1 . . . i.. .66) where Gc is the connected Green’s function. . y) δ δJ(y) : Z[J] Z[0] . . . . pn ) = G (p1 .e.

For the symmetry factor consider the examples (given in G. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar ﬁeld) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. = −i g . or calculating full Green’s functions external lines are treated as propagators.66 and 10. Note that in calculating amputated Green’s functions external lines are neglected. to be precise iLI vertices in momentum space are introduced. Diagrammar) in the case of the interaction terms g f (10. the factor 2 corresponds to the two-points Green’s function having two identical ends). vertices and external particle wave functions in that diagram. For the interaction terms in the lagrangian. which cancels the factor 1/2 in the quadratic piece. 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. Veltman. If problems arise go back to the deﬁning expression for the generating function in 10.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. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). but overall momentum conservation at a vertex is understood. At these vertices each line can be assigned a momentum.27.69) LI (φ) = − φ3 − φ4 . which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically diﬀerent connected diagrams.27. ’t Hooft and M. (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).

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

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

1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to .Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. 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 . + 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. 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. The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). 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 . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). 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.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . λ (∂µ Aµ )2 2 1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . i p p i ui (p) ui (p) ¯ vi (p) ¯ incoming fermion outgoing fermion incoming antifermion outgoing antifermion i p p i vi (p) In writing down the expressions for Feynman diagrams one has to be aware that the wave functions and the propagators have more than one component. λ 1 ∂ (10.

s3 )(−ie)γ µ u(k. s)γν u(k ′ . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ .s3 . ¯ q2 (10. t2 (10. s′ )γµ u(k.5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. s1 ) ¯ −igµν u(p′ . s2 ).77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s′ ) ¯ u s. To lowest order (∝ α = e2 /4π) only one diagram contributes. s1 ) u e2 u(p′ . where u ¯ L(m) µν = = 1 2 u(k ′ . s3 )γ µ u(k. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].Feynman diagrams for scattering amplitudes 100 10.s4 e4 (m) µν (M) L L .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .79) . s4 )(−ie)γ ν u(p.76) 1 .s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . 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 . s2 )|2 ¯ q2 (10.s2 .s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10. s4 )γµ u(p. s)¯(k.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.

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

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

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

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

. Consider the process a + b → c + d. pd ) = 4π s 4π 8π s 4π dt dt √ √ . The various possibilities are referred to as diﬀerent reaction channels. = (11.27) with m2 = m2 − p2 = m2 + p2 + p2 .22) (11. Consider the 1-particle inclusive case. there is usually a preferred direction.19) (11.25) dR(s. H1 + H2 → h + X. in which one particle is detected.21) (11. tmax min λcd /4s.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. a c with q = λab /4s and q ′ = being 0 or 180 degrees.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. 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 . Of course there are not only 2-particle channels. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). At high energies. A speciﬁc reaction channel starts contributing at the threshold value (Eq. usually many particles are produced.18) (11. 2s (11. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. pT = (px .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. py ). . Examples appear in → π− + p (11. (11.17) (11. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. Writing E = mT cosh y. 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 .4). pc . Instead of the scattering angle. pz = mT sinh y. The initial state. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. 11. in exclusive measurements all particles are detected. the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . E . The variable s is always larger than a c b d the minimal value (ma + mb )2 .23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .20) → π0 + n → π+ + π− + n → .. usually is a 2-particle state. In inclusive measurements no particles are detected in the ﬁnal state. An often used set of invariants are the Mandelstam variables. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. however.

tt = s. ut = u) = Mab→cd (s. The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . u). u) with s = (p1 + p2 )2 = (p3 + p4 )2 . t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. η = − ln(tan(θ/2)) = tanh−1 (cos θ). u). (11. uu = (pa − p¯)2 = s. b ut = (pa − pd )2 = u. uu = s) = Mab→cd (s.2 Crossing symmetry In the previous chapter. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. ¯ tt = (pa − p¯)2 = s. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. t = (p1 − p3 )2 = (p2 − p4 )2 . b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. ¯ tu = (pa − pc )2 = t. 11. t.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. t. which means that rapidity distributions dN/dy maintain their shape. This is known as crossing symmetry. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). (11.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. tu = s. t-channel and u-channel processes respectively.30) (11.28) which (only involving angles) is easier to determine. ¯ b (11.31) . c bd Mad→c¯(su = u. u = (p1 − p4 )2 = (p2 − p3 )2 . 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.

t and u. . 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 + . u) that represents physical amplitudes in the physical regions for three scattering processes. i. . To ﬁnd the physical regions one looks for the boundaries of cos θs = cos θt = cos θu = 1 λs λs ab cd 1 λt λt ac bd 1 λu λu ad cb 2 2 2 2 s(t − u) + (Ma − Mb )(Mc − Md ) . (11.32) σ= trf T · V ρa ρb va co . _ 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. T where V and T indicate the volume and the time in which the experiment is performed. . Nc /T indicates the number of particles c detected in the scattering process. 1 N . 2 2 2 2 t(s − u) + (Ma − Mc )(Mb − Md ) . .Scattering theory 107 One can also phrase it in the following way: one has a single analytic function Mab→cd (s.e. The proportionality factor has the dimension of area and is called the cross section. (consider for convenience the rest frame for the target. . t.) · Nb · ﬂux(a).) is deﬁned as the proportionality factor in Nc = σ(a + b → c + . . while the ﬂux of the beam particles a is trf ﬂux(a) = ρa · va . say b) the cross section σ(a + b → c + . Nb indicates the number of (target) particles b. This is a consequence of the constraint 2 2 2 2 s + t + u = Ma + Mb + Mc + Md . 2 2 2 2 u(t − s) + (Ma − Md )(Mc − Mb ) . which for a density ρb is given by Nb = ρb · V . shown below for the case of equal masses.

.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. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . (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.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). as expected. (11. . N 1 . The decay time for moving particles τ is related to the decay time in the rest frame of that particle (the proper decay time τ0 ) by τ = γ τ0 . the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. . σ= √ 2 λab T · V (11. T · V ρa (11. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. N and T · V are covariant.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. Using ρtrf = a (0) (0) (0) lab lab lab ρa · γa = ρa · Ea /ma (where ρa is the rest frame density) and va = plab /Ea we have a trf ρa ρb va = (0) ρa ρb 2 λab 4 ma mb (0) (0) or with ρa = 2 ma . .T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). 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 (11.34) i. dt (11.e. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .38) (in which we can calculate −iMf i using Feynman diagrams). (11.T V. pn ). it is.35) This quantity is not covariant.Scattering theory 108 Although this at ﬁrst sight does not look covariant. .36) (11.

p1 .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 . respectively.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s.44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. .Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). θ) in quantum mechanics. note the sign and cos θ = qi · qf ). . q f ). 8π s 2π 8π s . see 11. (11. θ). p1 . √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). .4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. θ) or M (q i . for which one has dσ/dΩ = |f (E. The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . t) λab 4π dt. . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . 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. The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. p1 . The result is N = |Mf i |2 dR(s. (11.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. pn ) |M |2 q 32π 2 m2 dΩ |M |2 .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . (11. (11.42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. pn ). .47) M (s. q n ). . (11. . . .41) (11. q n ) q n qn √ Mni (q i . . implies for the scattering matrix M . θcm ) √ q 8π s 2 2 (11. 2 λab and for instance for two particles dσ = q ′ M (s. θ) = −8π s ℓ (in analogy with the expansion for f (E. pn ). 11.43) dΩcm = π M (s. θ)|2 .e. for which the √ amplitudes squared have been calculated in the previous chapter. 16π 2 s (11. ˆ ˆ Inserted in the unitarity condition for M . . .

σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). In general a given channel has |Sℓ (s)| ≤ 1.50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . the total cross section can be determined. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni .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. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 .49) or Im Mℓ = q |Mℓ |2 .54) ℓ . n (11. (11.e. q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni . parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). Using σ= dΩ q ′ M (s.51) and for the case that this is the only channel (purely elastic scattering.53) The diﬀerence is the inelastic cross section. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. θcm ) √ q 8π s 2 . ℓ (11.52) From the imaginary part of M (s. 0). q ˆ i. (11. 2i q 2i q (11. (11.

It is important to note that the physical width of a particle is the imaginary part of the two-point vertex at s = M 2 . . σel = σinel . For an unstable particle one expects that U (t. 11. s − M 2 + iM Γ (11. however. (q Mℓ )ij (s) = −M Γi Γj . The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . but still real. 0)|2 = e−Γt .e.56) (again disregarding spin). η = 0.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. i. 0). For instance the tree-level calculation of Feynman diagrams at order e2 was real. 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. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. ∝ U (t.57) This shows that Γ is the width of the resonance. such that |U (t.e. We note that unitarity is generally broken in a ﬁnite order calculation. which is given by a sum of the partial widths into the diﬀerent channels. in which case. To check unitarity we need the amplitude at order e4 . it is straightforward to write down the partial wave amplitude. For the amplitude in a scattering process going through a resonance.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. This is what we will do to discuss decay widths. The prescription for the pole structure. The product of amplitudes is of order e4 . The quantity Γ is precisely the width for unstable particles. 0) ∝ e−i(E−iΓ/2)t . .58) . 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 .55) (note that we have disregarded spin). the time-evolution operator. . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. (11. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. i. . 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.

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

writing N (s) M1 (s) = 2 + iM Γ . a quantity which we will ¯ ¯ encounter in the next chapter (section 12. As discussed for unstable particles. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. cm |va − vb | in the center of mass system for two colliding particles a and b. Calculate now fπ . we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width.999 877 Γ and Γ(π − → e− νe ) = 0.2 Calculate the ﬂux factor. Exercise 11.2).000 123 Γ.4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. 4.Scattering theory 113 Exercise 11.8. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. Thus.3 Complete the calculation of the decay width and lifetime of the pion by including the appropriate phase space factors and using the calculation of the squared matrix element in Exc. ρcm ρcm a b 1 cm . Exercise 11. Compare the results with the experimental π − lifetime τ = 2. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) .

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

17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). but expresses them with respect to diﬀerent bases.e. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. 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. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). However.14) (12. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12.16. 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 . which connects two identical vectors. (12.The standard model the basis for x G and x+dx 117 G. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). (12. should be independent of such transformations.19) This gives rise to a (path dependent) phase in each point.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x.e. = . Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). In principle such a phase in a given point is not observable. -dx -dy µ µ dyµ µ dx For a vector. The relation in Eq. 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).16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). 12. i. (12. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0.15) ψ(x). requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). (12. i. ds which is solved by dψ(s) ds = dxµ . ds s dxµ ds′ P exp ig A (s′ ) ψ(0).

2. 12. we can disregard those ’perturbations’ and look at the ideal situation. in reality a not perfectly symmetric situation. usually being (slightly) prejudiced by impurities. Nature will choose one. 12. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect). In this case there are many possible groundstates (determined by a ﬁxed direction in space). a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components. the groundstate itself appears degenerate. .The standard model 118 where Gµν = (i/g)[Dµ .20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . Dν ]. As there can be one and only one groundstate. Nevertheless. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν . and equivalently Aµ can be considered as a pure gauge eﬀect. e. φ1 φ .e. Similarly as the deﬁnition of the covariant derivative the eﬀect is thus frame-independent and we have the transformation law Gµν → S(x)Gµν S −1 (x). Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . Only if this quantity is nonzero a physically observable eﬀect of Aµ exists. i>j which is rotationally invariant. 2 2 4 −V (φ) The potential V (φ) is shown in ﬁg.g. 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. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. If it is zero one has dxµ Aµ (x) = 0.2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. i. This characterizes spontaneous symmetry breaking. φ= 2 φ3 (12. 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. external magnetic ﬁelds.

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

. This means that they are operators that create states from the vacuum. 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. ψ −→ ei α·T ψ. Taking the derivative. it is easy to see that one gets ∂µ V µ = i ψ [M. and one must have mπa = 0. called ﬂavors. (12. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . ∂ ∂ (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′ .e. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. i. H] = 0.26) for all the states labeled by a corresponding to ’broken’ generators.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). Including a sum over the diﬀerent ﬂavors (up. A bit more formal. T = τ /2.32) (12.) one can write L = f ψ f (i/ − Mf )ψf . π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. i. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. Using the Dirac equation. if the generators Qa generate a symmetry.31) . Goldstone mode. etc. i. M = . .30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices.. strange. a ∂ µ Jµ = 0. [Qa . (12. ψu mu md ψ = ψd . Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). a massless Goldstone boson for each ’broken’ generator. 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.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. = ψ(i/ − M )ψ. (12. π (12.The standard model 120 by degenerate representations of the symmetry group. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. V µ = ψγ µ T ψ. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. .e. a and a′ are degenerate states. The generators Qa (in that case the rotation operators Lz . irrespective of the fact if they annihilate the vacuum. there must be a nonzero expectation value 0|Jµ (x)|π (k) . Neglecting at this point the local color symmetry.e. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). T ] ψ. . denoted |π a (k) . down. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).

in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR .e. Therefore the world of up and down quarks describing pions. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. chiral symmetry is not perfect. (leaving out the ﬂavor structure) the same as ψγ5 ψ. I. fπ = 0 remains. which is approximately true for the up and down quarks. is by nature not realized in the Weyl mode. i. T } γ5 ψ.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices.27. Note that these transformations also work on the spinor indices. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. T γ5 = τ γ5 /2.e. generated by TR/L = 1 (1 ± γ5 )T . as the quantity ∂µ Aµ (x). (12. The conclusion is that the original symmetry of the lagrangian is spontaneously broken and as the vector part of the symmetry is the well-known isospin symmetry. 12. Aµ = ψγ µ T γ5 ψ. socalled axial vector transformations. How can we see this. where the quark masses are not completely zero. with almost degenerate masses (Mp = 938. including in Dirac space a γ5 matrix. ψ −→ ei α·T γ5 ψ. ∂ ∂ (12.34) Using the Dirac equation.g. but the fact that the symmetry is not perfect and the right hand side of Eq. however. much smaller than any of the other mesons or baryons.e. parity would not play a role in the world. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. proton and neutron. The currents corresponding to this symmetry transformation are again found using Noether’s theorem.35 is nonzero. The chiral ﬁelds ψR and ψL are transformed into each other under parity. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. From the number of broken generators it is clear that one expects three massless Goldstone bosons. which is equivalent 2 to the V-A symmetry. if in ﬂavor space M is proportional to the unit matrix.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. There exists another set of symmetry transformations. This situation. i. for which the ﬁeld (according to the discussion above) has the same behavior under parity. etc. behaves as a pseudoscalar particle (spin zero. while the groundstate is only invariant under isospin rotations (R = L). 12. e. . nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . This combined symmetry is what one calls chiral symmetry. (12. (12. From group theory (Schur’s theorem) one knows that the latter can only be true.6 MeV/c2 ). This symmetry. the lagrangian density is invariant under left (L) and right (R) rotations independently. ψR/L = 1 (1 ± γ5 )ψ. a doublet (isospin 1/2) of nucleons. T ] = 0.36) If the mass is zero the lagrangian is split into two disjunct parts for L and R showing that there is a direct product SU (2)L ⊗ SU (2)R symmetry.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. but also a triplet (isospin 1) of pions. M = m · 1. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions.3 MeV/c2 and Mn = 939. is what happens in the real world. it is easy to see that one gets ∂µ Aµ = i ψ {M. i.e. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . etc. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . or.e. parity minus). We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. In the real world. both are very small. stated equivalently.

one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds.40) a Dµ φ = ∂µ φ − ig Wµ La φ. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. . made into a gauge theory. having two independent spin components) and three scalar ﬁelds (one degree of freedom each). as in this case there are no massless Goldstone bosons. Initially there are 3 massless gauge ﬁelds (each. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . 0 φ= (12.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν .The standard model 122 12. (12. again 9 degrees of freedom. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). . 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). thus 9 independent degrees of freedom. .39) (12. is made. One may wonder about the degrees of freedom. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. After symmetry breaking the same number (as expected) comes out.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).e. i. while the number of massive gauge bosons coincides with the number of ’broken’ generators. 0 0 0 = .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. 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 + . (12. 0 ϕc = 0|φ |0 = 0 . in this case three-dimensional) representation reads (La )ij = −i ǫaij . . The diﬀerence comes when we reparametrize the ﬁeld φ. We have the possibility to perform local gauge transformations. 4 2 where Since the explicit (adjoint.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 . 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). (12. like a photon.37) (12. The symmetry is broken in the same way as before and the same choice for the vacuum. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. but one has 1 massless gauge ﬁeld (2).

the quarks (up.49) Next the SU (2)W ⊗ U (1)Y symmetry is made into a local symmetry introducing gauge ﬁelds Wµ and Bµ in the covariant derivative Dµ = ∂µ − i g Wµ · TW − i g ′ Bµ YW /2. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.50) (12. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature.45) (12. The procedure. 2 2 = ∂µ R + i g ′ Bµ R. (12. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− .48) (12. i. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained.The standard model 123 12. etc. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. etc. however. left. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L.and righthanded particles. its corresponding neutrino (νe . under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. ψR/L = 2 (1 ± γ5 )ψ decouple. τ ). explicitly L −→ e−i β/2 L.51) . Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. (12.44) which has an SU (2)W symmetry under transformations eiα·TW . 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.e. 1 As they are massless. U(1)Y U(1)Y (12. νµ and ντ ) and their antiparticles. electromagnetic and weak forces. neutrinos. the leptons (elektrons. down. This is written as 3 Q = TW + YW . implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. while the lefthanded particles form a doublet.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. muons.47) and YW is considered as an operator that generates a U (1)Y symmetry.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. that gives a good description of the physical world. muon µ− or tau τ − ). = ∂µ L − (12. R. ∂ ∂ ∂ (12.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. 2 (12. R −→ e−i β R.

this leads to the lagrangian L (h1) . with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .58) (12.52) 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 . . .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. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. h) (12. . the U (1)Q symmetry (generated by the charge operator).57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . † i i g Wµ · τ − g ′ Bµ )φ.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. leading to the parametrization 1 0 φ(x) = √ (12. . which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. L where L (h1) (h) =L (h1) +L (h2) . (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. + 8 (12.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . 4 4 In order to break the symmetry to the symmetry of the physical world. For the classical ﬁeld the choice θ4 = v is made. . Putting this in leads to L (f ) =L (f 1) +L (f 2) .59) . (12. (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). Using local gauge invariance θi for i = 1.

.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .61) (12.63) (12. (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 .65) The weak mixing angle is related to the ratio of coupling constants.66) From the coupling to the photon.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 . we can read oﬀ e = g sin θW = g ′ cos θW . 2 Zµ (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .64) (12.67) The coupling of electrons or muons to their respective neutrinos.60) (12. 2 (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 . 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 + . . 2θ 2θ 4 cos W cos W 0. 4 2 MW g2 v2 = . 2 + g ′2 g 3 g ′ Wµ + g Bµ (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 . g ′ /g = tan θW .69) .

76) with CV CA 3 = TW − 2 sin2 θW Q.e.71) GeV −2 = = = = . . −5 (12.231 2. (12. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.75) The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with 3 TW 1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 . from the MW mass is about 250 GeV. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. g ′ and v determine a number of experimentally measurable quantities.e. v (12. Finally we want to say something about the weak properties of the quarks. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). (12. Comparing this with the total width into (invisible!) channels. The coupling to the Higgs particle is weak as the value for v calculated e. 1. giving rise to a running coupling constant after renormalization of the ﬁeld theory. i.40 GeV.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 91.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.166 4 × 10 0. g2 MZ 2 2 (CV + CA ). 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. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.77) (12.5 MeV (exp. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. muon or tau.g. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . me /v is extremely small. (12. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). 3 = TW .70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.80) First.72) (12. (12.19 GeV.73) (12. i.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. 2 2 2 (12. 80. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).74) (12.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. one has the relation g2 e2 1 GF √ = . With the choosen vacuum expectation value for the Higgs ﬁeld. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.

etc. As shown in Fig. suggesting an underlying larger unifying symmetry group. This is only true if the mixing matrix is at least three-dimensional.3: Appropriate YW and TW assignments of quarks. for which SU (5) or SO(10) are actually nice candidates. Comparing neutron decay and Λ decay one can get an estimate of the mixing parameter Vus in the socalled Cabibbo-Kobayashi-Maskawa mixing matrix. this requires particular YW -TW assignments to get the charges right. i. Q = TW + YW /2 and constant along speciﬁc diagonals as indicated in the ﬁgure. eR Q = −1 u d W eνe s W - u eνe where The quarks also turn out to ﬁt into doublets of SU (2)W for the lefthanded species and into singlets 3 for the righthanded quarks. In principle one complex phase is allowed in the most general form of the CKM matrix.81) n −→ pe− νe ⇐⇒ d −→ ue− νe . ¯ 3 This mixing allows all quarks with TW = −1/2 to decay into an up quark. which can account for the (observed) CP violation of the weak interactions. The pattern is actually intriguing. t c u ′ . ′ ′ 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. The electric charge Q is then 3 ﬁxed. their antiparticles and the electroweak gauge bosons as appearing in each family. 12.3. 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. Decay of B-mesons containing b-quarks allow estimate of Vub . We will not discuss this any further in this chapter. but with diﬀerent strength.e. CP violation requires three generations. Λ −→ pe− νe ¯ ⇐⇒ s −→ ue− νe . The .The standard model 127 YW eL uR νR +1 eR dL dL _ uL W 0 W −1/2 B0 +1/2 W + I3 W uR dR eL −1 dR Q = +1 νL Q=0 uL 3 Figure 12. leptons.

227. the weak mass eigenstates are L (h2.e. each of these containing the three families. 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. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12.5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2. i.34. UR = uR cR tR . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . (12. One then reads oﬀ that starting with the family basis as deﬁned via the left doublets that the mass eigenstates (and states coupling to the Higgs ﬁeld) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL .86) the unitary CKM-matrix introduced above in an ad hoc way. The Higgs ﬁeld is still limited to one complex doublet.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. † Λu Λ† = Vu G2 Vu . .g.82) space. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .82 and ρ ≈ 0. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . ms and mb ).q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. (12. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . respectively). e. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices.83) where Vu and Vd are unitary matrices.22 and η ≈ 0. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. u u where we include now the three lefthanded quark doublets in QL . 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 . d d (12. A ≈ 0. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues.

For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . since the weak current is the only place where they show up. 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 . (12. mµ and mτ . (12.e. UPMNS a parametrization that in principle also could have been used for quarks. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). As before.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. and obtain L (h2.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . The mass ﬁelds ER † mass † mass = We ER . i. 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 .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .5×10−3 eV2 . NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. we ﬁnd massless neutrinos.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . (12. EL = Ve† EL .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . In this case. = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (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 . (12.92) .ℓ) = −LφΛe ER − ER Λ† φ† L. e N L= L EL (12. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. there is no family mixing for massless neutrinos. 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 lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . one can write Λe = Ve Ge We and we ﬁnd † L (h2. decoupling from all known interactions.89) with three righthanded neutrinos added to the previous case.

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.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section.94) M L NL NL + M L NL NL . Thus (disregarding family structure) one has two Majorana neutrinos. the massless and massive Majorana neutrinos.95) 2 In order to have more than a completely decoupled sector.ν = − MR NR NR + MR NR NR . with for the righthanded sector a mass term MR . the left. − MD MR nR 2 (12. one must for the neutrinos as well as charged leptons. are equivalent to two decoupled Majorana neutrinos (see below). This is called the see-saw mechanism (outlined below). is to add in Eq. Thus 2 Υ ′ = nc + nL . a unitary matrix relating them to the weak eigenstates. 1 c ∗ c L mass. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. coupled by a Dirac mass term. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. 2 R ψ = nR + nL . 12. We use here the primes starting with the weak eigenstates.and righthanded species then just form a Dirac fermion. (12. For the leptons. The mass matrix can be written as ML |MD | eiφ M = (12. Absorption of phases in the states is not possible for Majorana neutrinos.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. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. For these light Majorana neutrinos one has.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. one being massive.100) |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2 1 nc nR M L M D nL L c + h. 1 L Υ ′ = nR + nc .ν =− containing three (CP-violating) phases (α1 . 1 ∗ c c (12. hence the mixing matrix becomes iα /2 0 0 e 1 (12.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos.98) (12. 0 0 1 L mass.99) . The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors.c. For the neutrino sector. as above. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). Actually. couple the right. (12. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . however. α2 and δ). 2 although this option is not attractive as it violates the electroweak symmetry.

e. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . i ψ {M. 2 U (M M † ) U † = M0 . a and a′ are degenerate states.102) = 1 2 2 ML + MR + 2|MD |2 2 ± 2 2 2 2 (ML − MR )2 + 4|MD |2 (ML + MR + 2ML MR cos(2φ)) . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families). Υ2R nR (12. i. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. related via c Υ1R = U nL .e. Exercise 12. .The standard model 131 taking ML and MR real and non-negative. H] = 0.105) for i = 1. (12. [Qa .103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . (12. 2 with real masses Mi . Exercise 12. The diagonalization is analogous to what was done for the Λ-matrices and one ﬁnds U M U T = M0 with a (unitary) matrix U . Exercises Exercise 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 .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 . i. Prove that if the generators Qa generate a symmetry.1 Consider the case of the Weyl mode for symmetries. T ] ψ. T } γ5 ψ.2 Derive for the vector and axial vector currents.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 . (12.

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

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

relativistic quantum mechanics

relativistic quantum mechanics

Are you sure?

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

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

Get the full title to continue

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

scribd