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

# QUANTUM FIELD THEORY

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

November 2011 (version 6.04)

Contents

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

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

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

1

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

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

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

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

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

Diagrammatica.3 6.2 2. Peskin & Schroeder and Srednicki.3 4.3 2. I refer to the book of Srednicki [1] and Ryder [2].7 2. Cambridge University Press.1 2. Quantum Field Theory.5 4.2 3.4 2. 1.2 6.3 2. 1980. 6. 5. 3. 1995. Cambridge University Press.3 3. Itzykson and J. Veltman. 1985. 4. Cambridge University Press.4 3.1 3.4 2.1 1. M. Other text books of Quantum Field Theory that are useful are given in refs [3-6].1 2.5 2. 3. 2007. The notes contain all essential information. Weinberg. 1995.E.1 3.2 2.3.2 3.4 3. M.2 3.4 6.3 3.2 4. These notes Section 1. Quantum Field Theory. II: Modern Applications. Addison-Wesly.3 3. Cambridge University Press. 1994.2 2.1. Corresponding chapters in books of Ryder. 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). I: Foundations. 3. L. An introduction to Quantum Field Theory.4 4. M. Srednicki.2 3. Quantum Field Theory.1 6.4 3. Zuber.2 3. Peskin and D. Vol. Vol. The quantum theory of ﬁelds.-B.2 6. 2.1 5. Ryder.V.H. Schroeder. C.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.3 2. McGraw-Hill.2 1.5 2. 1996. S.1 4.3 3.1 2 2 3.3 2.3 References As most directly related books to these notes.3 2. Cambridge University Press. but are rather compact. 2.2.2 36 22 .5 5.3 3.8 3.

3 8.5 4.3 11.3 3.4 8.6 3.3 6.1 9.5 6 7 6 6.2 8. 6.4 12.3 20.6 3.4 9. 8.5 9.2 11.5 2. 6.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.10 8.6 10.8 5 4.5 .1 10.3 7.5 7.7 4. 6.3 2.5 9.1 8.8 6.5 6.1 7. 2.1 20.6 8.1 20.5.2.3 7.4. 4.3 10.2.1 20.1.2 7.6 9.2 4.3.2 10.1.5 4.3 9.4. 6. 4.3.3 8.2 5. 4. 20.3 6 8 9 Srednicki 3 3 3 37 5.1 5.6 3.3.7 5. 9.4. 9.2.2 8.2 12.1 9.1.4 7.4 12.5 7.2.6 4.7.1 11.4 10.3 12.5 11. 6. 2.4 4.4 2.2 9. 4.4 3.6 6.1 12.4 These notes Section 7.

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 6.2 .

4) where δij is the Kronecker δ function. c = 299 792 458 m s−1 . Indeed. They satisfy the well-known (canonical) operator commutation relations [ri .3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. It is given by pop ψ(r) = −i ∇ψ(r). pj ] = i δij .582 119 15 (56) × 10−22 MeV s.1) In the limit that the action S is much larger than . quantum eﬀects do not play a role anymore and one is in the classical domain.054 571 68 (18) × 10−34 J s = 6.6) .2) In the limit that v ≪ c one reaches the non-relativistic domain. (1. (1. Corresponding position and momentum operators do not commute. (1. 1 (1. One can talk about states |r and the wave function ψ(r) = r||ψ . In quantum mechanics. (1. ˜ ψ(p) = i d3 r exp − p · r ψ(r). the position can in principle be determined at any time with any accuracy. ≡ h/2π = 1. In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-deﬁned concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. being eigenvalues of the position operators. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). In quantum mechanics a special role is played by Planck’s constant h.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. the action of the momentum operator in the coordinate representation is not as simple as the position operator. S ≫ . ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. (1.Chapter 1 Introduction 1. In special relativity a special role is played by the velocity of light c. usually given divided by 2π.

Introduction

2

and ψ(r) =

d3 p exp (2π )3

i

˜ p · r ψ(p).

(1.7)

The existence of a limiting velocity, however, leads to new fundamental limitations on the possible measurements of physical quantities. Let us consider the measurement of the position of a particle. This position cannot be measured with inﬁnite precision. Any device that wants to locate the position of say a particle within an interval ∆x will contain momentum components p ∝ /∆x. Therefore if we want ∆x ≤ /mc (where m is the rest mass of the particle), momenta of the order p ∝ mc and energies of the order E ∝ mc2 are involved. It is then possible to create a particle - antiparticle pair and it is no longer clear of which particle we are measuring the position. As a result, we ﬁnd that the original particle cannot be located better than within a distance /mc, its Compton wavelength, ∆x ≥ mc . (1.8)

For a moving particle mc2 → E (or by considering the Lorentz contraction of length) one has ∆x ≥ c/E. If the particle momentum becomes relativistic, one has E ≈ pc and ∆x ≥ /p, which says that a particle cannot be located better than its de Broglie wavelength. Thus the coordinates of a particle cannot act as dynamical variables (since these must have a precise meaning). Some consequences are that only in cases where we restrict ourselves to distances ≫ /mc, the concept of a wave function becomes a meaningful (albeit approximate) concept. For a massless particle one gets ∆x ≫ /p = λ/2π, i.e. the coordinates of a photon only become meaningful in cases where the typical dimensions are much larger than the wavelength. For the momentum or energy of a particle we know that in a ﬁnite time ∆t, the energy uncertainty is given by ∆E ≥ /∆t. This implies that the momenta of particles can only be measured exactly when one has an inﬁnite time available. For a particle in interaction, the momentum changes with time and a measurement over a long time interval is meaningless. The only case in which the momentum of a particle can be measured exactly is when the particle is free and stable against decay. In this case the momentum is conserved and one can let ∆t become inﬁnitely large. The result thus is that the only observable quantities that can serve as dynamical coordinates are the momenta (and further the internal degrees of freedom like polarizations, . . . ) of free particles. These are the particles in the initial and ﬁnal state of a scattering process. The theory will not give an observable meaning to the time dependence of interaction processes. The description of such a process as occurring in the course of time is just as unreal as classical paths are in non-relativistic quantum mechanics. The main problem in Quantum Field Theory is to determine the probability amplitudes between well-deﬁned initial and ﬁnal states of a system of free particles. The set of such amplitudes p′ , p′ ; out|p1 , p2 ; in ≡ p′ , p′ ; in|S|p1 , p2 ; in determines the scattering matrix or S-matrix. 1 2 2 1 Another point that needs to be emphasized is the meaning of particle in the above context. Actually, the better name might be ’degree of freedom’. If the energy is low enough to avoid excitation of internal degrees of freedom, an atom is a perfect example of a particle. In fact, it is the behavior under Poincar´ transformations or in the limit v ≪ c Gallilei transformations that determine the description e of a particle state, in particular the free particle state.

1.2

Units

It is important to choose an appropriate set of units when one considers a speciﬁc problem, because physical sizes and magnitudes only acquire a meaning when they are considered in relation to each other. This is true speciﬁcally for the domain of atomic, nuclear and high energy physics, where the typical numbers are diﬃcult to conceive on a macroscopic scale. They are governed by a few fundamental units and constants, which have been discussed in the previous section, namely and

Introduction

3

c. By making use of these fundamental constants, we can work with less units. For instance, the quantity c is used to deﬁne the meter. We could as well have set c = 1. This would mean that one of the two units, meter or second, is eliminated, e.g. given a length l the quantity l/c has the dimension of time and one ﬁnds 1 m = 0.33 × 10−8 s or eliminating the second one would use that, given a time t, the quantity ct has dimension of length and hence 1 s = 3 × 108 m. Table 1.1: Physical quantities and their canonical dimensions d, determining units (energy)d . quantity time t length l energy E momentum p angular momentum ℓ mass m area A force F charge (squared) e2 Newton’s constant GN velocity v quantity with dimension energyd t/ l/( c) E pc ℓ/ mc2 A/( c)2 F c α = e2 /4πǫ0 c GN /( c5 ) v/c dimension (energy)−1 (energy)−1 (energy)1 (energy)1 (energy)0 (energy)1 (energy)−2 (energy)2 (energy)0 (energy)−2 (energy)0 canonical dimension d -1 -1 1 1 0 1 -2 2 0 -2 0

In ﬁeld theory, it turns out to be convenient to work with units such that and c are set to one. All length, time and energy or mass units then can be expressed in one unit and powers thereof, for which one can use energy (see table 1.1). The elementary unit that is most relevant depends on the domain of applications, e.g. the eV for atomic physics, the MeV or GeV for nuclear physics and the GeV or TeV for high energy physics. To convert to other units of length or time we use appropriate combinations of and c, e.g. for lengths c = 0.197 326 968 (17) GeV fm or for order of magnitude estimates c ≈ 0.2 GeV fm = 200 eV nm, implying (when 1 fm = 10−15 m ≈ 5 GeV−1 . For areas, e.g. cross sections, one needs

2 2

(1.9) = c = 1) that (1.10)

c

= 0.389 379 323 (67) GeV2 mbarn

(1 barn = 10−28 m2 = 102 fm2 ). For times one needs = 6.582 119 15 (56) × 10−22 MeV s, (1.11)

implying (when = c = 1) that 1 s ≈ 1.5 × 1024 GeV−1 . Depending on the speciﬁc situation, of course masses come in that one needs to know or look up, e.g. those of the electron or proton, me mp = 9.109 382 6 (16) × 10−31 kg = 0.510 998 918 (44) MeV/c , = 1.672 621 71 (29) × 10

−27 2

(1.12) (1.13)

kg = 0.938 272 029 (80) GeV/c .

2

Furthermore one encounters the strength of the various interactions. In some cases like the electromagnetic and strong interactions, these can be written as dimensionless quantities, e.g. for electromagnetism the ﬁne structure constant α= e2 = 1/137.035 999 11 (46). 4π ǫ0 c (1.14)

Introduction

4

For weak interactions and gravity one has quantities with a dimension, e.g. for gravity Newton’s constant, GN = 6.708 7 (10) × 10−39 GeV−2 . (1.15) c5 By putting this quantity equal to 1, one can also eliminate the last dimension. All masses, lengths and energies are compared with the Planck mass or length (see exercises). Having many particles, the concept of temperature becomes relevant. A relation with energy is established via the average energy of a particle being of the order of kT , with the Boltzmann constant given by k = 1.380 650 5(24) × 10−23 J/K = 8.617 343(15) × 10−5 eV/K. (1.16)

Quantities that do not contain or c are classical quantities, e.g. the mass of the electron me . Quantities that contain only are expected to play a role in non-relativistic quantum mechanics, e.g. the Bohr radius, a∞ = 4πǫ0 2 /me e2 or the Bohr magneton µe = e /2me . Quantities that only contain c occur in classical relativity, e.g. the electron rest energy me c2 and the classical electron radius re = e2 /4πǫ0 me c2 . Quantities that contain both and c play a role in relativistic quantum mechanics, e.g. the electron Compton wavelength −e = /me c. It remains useful, however, to use λ and c to simplify the calculation of quantities.

1.3

**Conventions for vectors and tensors
**

3

**We start with vectors in Euclidean 3-space E(3). A vector x can be expanded with respect to a basis ei (i = 1, 2, 3 or i = x, y, z), ˆ x=
**

i=1

xi ei = xi ei , ˆ ˆ

(1.17)

to get the three components of a vector, xi . When a repeated index appears, such as on the right hand side of this equation, summation over this index is assumed (Einstein summation convention). Choosing an orthonormal basis, the metric in E(3) is given by ei · ej = δij , where the Kronecker delta ˆ ˆ is given by 1 if i = j δij = . (1.18) 0 if i = j, The inner product of two vectors is given by x · y = xi yi ei · ej = xi yj δij = xi yi . ˆ ˆ (1.19)

The inner product of a vector with itself gives its length squared. A vector can be rotated, x′ = Rx or x′ = Rij xj leading to a new vector with diﬀerent components. Actually, rotations are those real, i linear transformations that do not change the length of a vector. Tensors of rank n are objects with n components that transform according to Ti′1 ...in = Ri1 j1 . . . Rin jn Tj1 ...jn . A vector is a tensor of rank 1. The inner product of two vectors is a rank 0 tensor or scalar. The Kronecker delta is a constant rank-2 tensor. It is an invariant tensor that does not change under rotations. The only other invariant constant tensor in E(3) is the Levi-Civita tensor if ijk is an even permutation of 123 1 −1 if ijk is an odd permutation of 123 ǫijk = (1.20) 0 otherwise.

Vectors are denoted x = xµ nµ . such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 .26) (1..Introduction 5 that can be used in the cross product of two vectors z = x × y. however. x2 . x1 . linear transformations that do not change the length of a four-vector are called the Lorentz transformations. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . The (invariant) lengths often have special names. ˆ ˆ (1. in which case zi = ǫijk xj yk . . denoted as V ′µ = Λµν V ν . x3 ) = (t. The Poincar´ transformations include e e Lorentz transformations and translations.3) with basis nµ (µ = ˆ 0. δim δjm δkm δin δjn δkn . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. the points lie on the lightcone and they can be connected by a light signal. Because of the diﬀerent signs occurring in gµν . (1.25) Within Minkowski space the real. Λµn n T ν1 . it is convenient to distinguish lower indices from upper indices. x2 . we will use upper indices for the three-vectors. xi are the three space components.24) The quantity gµν ≡ nµ · nν is the metric tensor.21) (1.. When 3-dimensional space is considered as part of Minkowski space. (1.3). xµ = gµν xν . One has x2 = xµ xµ = t2 − x2 . No diﬀerence is made between upper or lower. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. x3 ). x) = (x0 . In special relativity we start with a four-dimensional real vector space E(1. The lower indices are constructed in the following way. These transformations do change the components of a vector. So we could have used all upper indices in the above equations.µn = Λµ1ν1 . . one must distinguish tensors with upper or lower indices and one can have mixed tensors. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . Unlike in Euclidean space. • ds2 < 0 (spacelike intervals). • ds2 = 0 (lightlike or null intervals). One can distinguish: • ds2 > 0 (timelike intervals). x1 . The transformations that leave invariant the distance ds2 = dt2 − (dx2 + dy 2 + dz 2 ) between two points are called inhomogeneous Lorentz transformations or Poincar´ transformations. 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 .. independent of a coordinate system. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero).νn can be conν structed. The length (squared) of a vector is obtained from the scalar ˆ product.22) (1. and are given by (x0 . Many other four vectors and tensors transforming like T ′ µ1 . The real. −x). x2 = x · x = xµ xν nµ · nν = xµ xν gµν .1..23) δjm δkn − δjn δkm . where the coordinate t = x0 is referred to as the time component. are . the invariant length or distance (squared) is not positive deﬁnite.2. 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. We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices. 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 . Relations relating tensor expressions. In Minkowski space.

. p). The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . explicitly written as (p0 .32) . ∂3 ) = ∂ ∂ ∂ ∂ . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 .29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . (1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.Introduction 6 called covariant. deﬁned by ∂2 − ∇2 . ∂1 .30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). (1. because one has (1. = ∂ .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. (1. = −24. Each of these permutations leads to a minus sign. not aﬀected by Lorentz transformations. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. It is an invariant tensor. . (1. starting from ǫ0123 = 1.34) (1.28) ν ν Note that gµ with one upper and lower index. ∂2 . p) = (E. (1. but more important has the same eﬀect on lefthandside and righthandside. ′ ′ . 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.∇ . constructed via the metric tensor itself. where m is called the mass of the system. The length squared of ∂ is the d’Alembertian operator. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α.36) Exercises Exercise 1. ∂t (1.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. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . Note that in this equation one has on left.33) (1. The product of two epsilon tensors is given by g µµ ′ g νµ ′ g ρµ ′ g σµ g νν ′ g ρν ′ g σν ′ ′ ′ ǫµνρσ ǫµ ν ′ ′ ′ ρ σ′ = − g µν ′ g νν ′ g ρν ′ g σν g νρ ′ g ρρ ′ g σρ ′ ′ ′ g µρ ′ g νρ ′ g ρρ ′ g σρ g νσ ′ g ρσ ′ g σσ ′ ′ ′ g µσ ′ g νσ ′ g ρσ ′ g σσ . ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . a∞ = 4πǫ0 2 /me e2 .27) It is easy to convince oneself of the nature of the indices in the above equation. g0 = g1 = g2 = g3 = 1.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius.

a2 . . Also construct and calculate the Planck length Lpl .511 MeV. ν. n1 . 2. g+− and g−+ . n2 . (b) The coordinates (a0 . σ} is a permutation of {0. [Note: what is the MKS unit for V/T?] Exercise 1. ˆ ˆ .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). a1 . ρ.2 Prove the identity A × (B × C) = (A · C) B . a2 ] or [a− . n+ . g−− . a+ . 1. Exercise 1.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. ˆ ˆ ˆ ˆ These are n1 = (0. a+ .3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . me . 0. 3} the index α can only be equal to one of the indices in ǫµνρσ . c. n1 . Exercise 1. . 1. 0. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . e. This demonstrates how a careful use of units can save a lot of work. 0) ˆ n3 = (0.13 to give its actual value in kg. but only appropriate combinations. 1. (e) Note that /m = − c and use this to calculate in a simple way (avoiding putting in the value λ of e) the Bohr magneton µe = e /2me and the nuclear magneton µp = e /2mp in electronvolts per Tesla (eV/T). 1) ˆ . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. 0. by a simple few-line reasoning [For instance: If {µ.3. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . which is the Compton wavelength for the Planck mass. a+ . 0. 0. 1. . ]. a1 . a3 ) are the expansion coeﬃcients using the basis vectors n0 . (d) Use the value of the gravitational constant GN / c5 = 6. 0) ˆ n0 = (1. Also for the coordinates [a− . n3 . How are a+ and a− related to a+ and a− . n2 . ǫ0 . 0) ˆ n2 = (0. α and me c2 = 0.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). a2 ] we can ﬁnd basis vectors n− . re = e2 /4πǫ0 me c2 to the Compton wavelength. One does not need to know .(A · B) C using the properties of the tensor ǫijk given in section 1. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . 0. Compare it with the proton mass and use Eq. a1 .

7) k 0 = ± k2 + M 2 = ±Ek . 8 . t) = − ψ(r. But the replacement 2. ∂t 2M (2. ∂t2 (2.2) Acting on the wave function one ﬁnds for a free particle. E = p2 . Although it is straightforward to ﬁnd the solutions of this equation. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. p). the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite. one obtains (2. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. 2 + M 2 φ(r.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . t) = ∂2 − ∇2 + M 2 φ(r. t) = exp(−i k 0 t + i k · r). 2M (2.1 and 2. t) = 0. φk (r. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.1 The Klein-Gordon equation In this chapter. (2.2. namely plane waves characterized by a wave number k.3 are not covariant.4) where M is the particle mass. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. ∂t p −→ pop = −i∇.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions.5) with (k 0 )2 = k2 + M 2 .3) Equations 2.Chapter 2 Relativistic wave equations 2. p2 = pµ pµ = E 2 − p2 = M 2 . written as pµ −→ i∂µ is covariant (the same in every frame of reference). (2. (2. i ∂ ∇2 ψ(r. t).

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

2. x ∂µ ∂ µ + M 2 φ(˜) = 0. (2π)3 2Ek (2. Since a · b = a · ˜ it is easy to see that ˜ b. changing the sign of the spatial coordinates. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. Performing some Lorentz transformation x → x′ = Λx. for which we consider space inversion. (2π)3 2Ek (2. This shows how parity transforms k-modes into −k modes. −x) ≡ (˜µ ).20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k).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).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) .3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. which implies (xµ ) = (t. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) .e.6 We will explicitly discuss the example of a discrete symmetry.16) The consequence of this is discussed in Exercise 2. i. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes).Relativistic wave equations 10 In Eqs 2. x) → (t. φP (x) ≡ φ(˜) (P for parity).23) .15 one has elimated k 0 and in both equations k · x = Ek t − k · x. (2. Another symmetry is found by complex conjugating the KG equation.14 and 2. 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) .18) (2. (2. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. after restricting the energies to be positive).22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).17) (2. (2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). (2π)3 2Ek (2.

that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). t) ≡ d3 x φ(x.e.2 Show that if φ and φ∗ are solutions of the KG equation. (2π)3 2Ek where Ek = Exercise 2. d3 k F (Ek . ˜ b(k) = ei Ek t i ∂0 φ∗ (−k.Relativistic wave equations 11 i. ↔ ↔ Exercise 2.3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . k) = (2π)4 k2 + M 2 . t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ). Q = 0. QV ≡ satisﬁes ˙ QV = − ds · j.1 Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume. |f ′ (xn )| .15. letting V → ∞. is conserved. Exercise 2. S V d3 x j 0 (x). This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. k).4 ˜ Write down the 3-dimensional Fourier transform φ(k. ˙ and thus for any normalized solution the full ‘charge’. Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. t). Note that a(k) and b(k) are independent of t. t) = ei Ek t (i∂0 + Ek ) φ(k. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k). t). For the real ﬁeld one has aC (k) = a(k). t) for the mode expansion in Eq. 2. Exercises Exercise 2.

Exercise 2. (c) Similarly. B] = AB − BA and the anticommutator as {A. i. Note that a{µ bν} indicates symmetrization. and β 2 = I.5 (a) As an introduction to parts (b) and (c). ∂ ψ(r. αj } = 2 δ ij I. 2. j 0 = ψ†ψ and j = ψ † αψ. 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). {αi . t) = (−i α · ∇ + m β) ψ(r. show that for two ﬁelds φ1 (with coeﬃcients a1 and b∗ ) 1 and φ2 (with coeﬃcients a2 and b∗ ) one has 2 d3 x φ∗ (x) φ2 (x) 1 = d3 k 2 (2π)3 4Ek a∗ (k)a2 (k) + b1 (k)b∗ (k) 1 2 + a∗ (k)b∗ (−k) e2iEk t + b1 (k)a2 (−k) e−2iEk t . making sure that each component of ψ satisﬁes the KleinGordon equation (in terms of diﬀerential operators ∂µ ∂ µ +m2 = 0) requires the anticommutation relation2 {αi . We will encounter these quantities later as energy and momentum. d3 x (∂ {0 φ)∗ (∂ i} φ). From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. 1 2 (b) Express now the full charge QV (t) (exercise 2. t). also using the result in (a). β} = 0. .15. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write. i (a) Show that relativistic invariance.6 Write down the mode expansion for the Lorentz transformed scalar ﬁeld φ′ (x) and show that it implies that the Lorentz transformed modes satisfy a′ (Λk) = a(k). Exercise 2.1) for a complex scalar ﬁeld current (exercise 2. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles).Relativistic wave equations 12 Exercise 2. ∂t where ψ is a wave function and the nature of α and β is left open for the moment. Is the time dependence of the result as expected? Note that the mode amplitudes will become creation and annihilation operators after quantization of the ﬁelds (Chapter 7). express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . in terms of the a(k) and b(k).2) in terms of the a(k) and b(k) using the expansion in Eq.e. a{µ bν} ≡ aµ bν + aν bµ .7 To solve the problem with positive and negative energies and get a positive deﬁnite density. B} = AB + BA.

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

7) = a0 0 −1 i 0 1 0 0 1 • [x.5) Rotations in general do not commute. 2 (3.2 π) (-n. is a 3-dimensional Lie-group. One way of viewing the parameter space. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). y]] = 0 (Jacobi identity).8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. Ly and Lz . π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. ] that satisﬁes • ∀ x. with real a’s and 3 (ai )2 = 1.4) R . (3. ˆ therefore. and thus (σ · n)2 = 1. ˆ R(ϕ. namely that there exists a bilinear product [. [z. z ) = lim N →∞ that are generated by i 0 . which reﬂects itself in the noncommutation of the generators. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . • [x. [x. x]] + [z. Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ .z N N or (Lk )ij = −i ǫijk .6) Summarizing. Locally this is a 3-dimensional Euclidean space and SU (2). These generators form a threedimensional Lie algebra SO(3). In the same way we can consider rotations around the x. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations.and y-axes 0 0 0 0 0 0 0 −i . . (3. π) (n. Lj ] = i ǫijk Lk . y] = −[y. They satisfy the commutation relations [Li .Groups and their representations 14 (n. Next. x] = 0 (thus [x. 0 (3. y] ∈ A. y ∈ A ⇒ [x. z]] + [y. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. for rotations around z for instance. 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. [y. x]).1: the parameter spaces of SO(3) (left) and SU (2) (right).

with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. immediately sees that the Lie algebras are identical. H] = 0 for g ∈ G.10) σ ˆ · n. In the full parameter space. it is suﬃcient that the generators commute with H. (3. Suppose we have a system described by a Hamiltonian H. i. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. . n! 2! n=0 ∞ (3. SU (2) ≃ SO(3). The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. . n) i ∂ϕ = ϕ=0 (3.9) The parameter-space of SU (2).15) [g. They satisfy σi σj σk = i ǫijk . however.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. in particular that AB → RAB = RA RB ).Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + .e. 2! . for ﬁxed n. i. For instance in a function space translations. i. . φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + .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. also can be considered as a ﬁlled 3-sphere. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. thus. 3. n) ˆ −→ R(2π − ϕ. . since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. n) −→ R(ϕ. thus. Near the identity.14) For a Lie group. (3. n) π ≤ ϕ ≤ 2π.12) One. ˆ ˆ A(ϕ. 2 2 2 (3. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure).11) Thus σx /2. .e. sometimes in more than one way (but this will be discussed later). the above mapping corresponds to the trivial mapping of the Lie algebras. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). . [g. (3. H] = 0 for g ∈ G. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. n) ≈ 1 + i ϕ J · n. . 2 (3. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I.e.

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

the extension from a local one must be unique. The matrix elements of these unitary representations are known as D-functions. j. the conjugate transformation U ∗ acts on χ∗ . 1 |1.19) (3. then ǫχ → ǫσχ = ǫσǫ−1 ǫχ = −σ∗ ǫχ. ǫy . This works for SU (2). θ. all states in the representation space have the same energy. χ)|j. (j) (j) ′ The spin matrices for that Cartesian basis {ǫx . For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1. or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . the conjugate representation is not a new one. For an element in the group G the corresponding point in the parameter space can be reached in two ways. 2 From the (hermitean) representation Φ(g) of G. If rotations leave H invariant. e. Jz = i 0 i 0 −i 0 0 0 −i 0 0 0 0 0 . Jy = 0 0 0 . For those global representations. is not simply connected.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 . UE (ϕ. Inﬁnitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. the possibility thus exist that some extensions will not be well-deﬁned representations.e. Consider the j = 1/2 representation of SU(2).20) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles. θ. 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) .3). m′ |UE (ϕ. however. J = 0 0 + 0 0 0 0 0 0 0 0 √ 2 0 √ 0 . m = Dm′ m (ϕ. Of all groups of which the Lie algebras are homeomorphic. For SO(3). For SU(2). If a transformation U acts on χ. Given a representation. dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m . the topological structure of the group is important. SU (2) is the covering group of SO(3). 2 |1. For a decent (welldeﬁned) global representation. contractable and paths that run from a point at the surface to its antipode. −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . one can look at the conjugate representation. to be precise the states ǫχ transform via −σ ∗ . the simply connected group is called the (universal) covering group. any point in the parameter space can be reached in a unique way and any local (inﬁnitesimal) representation can be extended to a global one. (j) (3. if χ → σχ. however. θ. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2. J− = 2 0 0 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0 . that is any closed curve in the parameter space can be contracted to a point. . Because there exists a matrix ǫ such that −σ ∗ = ǫ σǫ−1 one immediately sees that appropriate linear combinations of conjugate states. however. i. one obtains the unitary representations Φ(g) = exp(i Φ(g)) of the full group G. For the transformations A in SU (2) and R in SO(3) these are commonly parametrized with Euler angles. ǫz } are 0 0 0 0 0 i 0 Jx = 0 0 −i . Explicitly. There exist two diﬀerent types of paths. If the group is simply connected. The group SO(3). χ) = eim ϕ dm′ m (θ) e−imχ .g. This turns out to be the case for all half-integer representations of the Lie algebra. 0 ≡ ǫ0 ≡ ǫz . however. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. χ) ≡ e−iϕJz e−iθJy e−iχJz .

σ/2 in a two-dimensional (spin 1/2) case. (3. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. (ii) |Λ00 | ≥ 1.g. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.Groups and their representations 18 3.27) (3. 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 . explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . From this property.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index.26) (3. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . (det Λ = +1 is called proper.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. We considered the generators and looked for representations in ﬁnite dimensional spaces.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . G = 10 (3. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. det Λ = −1 is called improper). Writing x as a column vector and the metric tensor in matrix form. consisting of the Lorentz group and translations. e.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. Using (i) and (ii) the Lorentz transformations can be divided into 4 classes (with disconnected parameter spaces) det Λ +1 +1 -1 -1 Λ00 ≥1 ≤ -1 ≥1 ≤ -1 (3. In this section we consider the Poincar´ group. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = .25) (3. it is convenient to use a vector notation for the points in Minkowski space. e To derive some of the properties of the Lorentz group. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.22) (Note that xT is a row vector).

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

g. (3. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. We will just require Poincar´ invariance for the generators themselves. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. 3 30 with η = ω and K = M (e. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . (3. We therefore ﬁnd (again) that there are six generators for the Lorentz group.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . K = M ).39) (e. inﬁnitesimally approximated by (I + ω. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ .38) Explicitly. In order to ﬁnd the commutation relations including the translation generators. The extension to the Poincar´ group is e ↑ straightforward. P+ etc. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. we continue using covariance arguments. i i0 i i0 1 ijk M jk 2ǫ (3. Also this group can be divided into four parts. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . There are six generators. [M µν . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . Summarizing. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). a). We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. Of the four parts only L↑ forms a group.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 .g.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. inf (3.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . ǫ).33) These four transformations form the Vierer group (group of Klein). three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). This is a + normal subgroup and the factor group is the Vierer group.36) (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. 3. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3.

Explicitly. check that U1 U2 corresponds with Λ1 Λ2 ). H] = i P i . J j ] = i ǫijk J k .Groups and their representations 21 inﬁnitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. known from non-relativistic quantum mechanics are obtained. P j ] = [P i . the generators J± in the case of the rotation group. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. a)P µ U † (Λ. that commute e among themselves. the generators M αβ no longer exclusively act in Minkowski space. For instance. a) = Λ−1 µ P ν . (3. other states in that representation are obtained by the action of operators outside the maximal commuting set. one obtains 2 [P i .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. P µ |p. .44) (3.42) (to decide on where the inverse is. Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . .38 could have been obtained in the same way. P j ] = i δ ij H/c2 . = pµ |p. 3. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. P ) in terms of the Hamiltonian and the three-momentum operators. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . . = −i (g µρ P ν − g νρ P µ ) . P ν ] [M µν . P j ] = i ǫijk P k . We have here reinstated c. a)P µ U (Λ. ν (3. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . the generators J 2 and J 3 in the case of the rotation group. . The eigenvalues of these will be the four-momentum pµ of the state. [J i .41) 2 At this point. . (3. 2 2 (3. For instance.47) (3. . K j ] = i ǫijk K k . They just state that M µν transforms as a tensor with two Lorentz indices. P ρ ] = 0. Inﬁnitesimally. [P µ . [J i . These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. as part of the set. writing the generator P = (H/c.43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν .45) Note that the commutation relations among the generators M µν in Eq. leads to U (Λ. H] = 0. (3. H] = [J i . The generators of the translations are the (momentum) operators P µ . Taking any one of the states in an irreducible representation. (3. 3. These form a group called the little group associated with that four vector.46) [K i . K j ] = −i ǫijk J k /c2 . [K i . a) = Λµν P ν or U † (Λ. [J i . [K i .48) . ǫ) = 1 − ωαβ M αβ + i ǫα P α .

2 exp (−isµ W µ ) |p. the third case represents the vacuum. 2 (3. 0. = = i exp − ǫµνρσ M µν pρ sσ |p. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . .52) (3.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. Pν ] = [Wµ .49) (3.51) The following properties follow from the fact that W µ is a four-vector by construction.56) .55) commute with all generators and therefore are invariants under Poincar´ transformations. . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. .50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . while the others have no physical signiﬁcance. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).54) This will enable us to pick a suitable commuting ’spin’ operator. 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. Wν ] = −i (gµα Wν − gνα Wµ ) . . 0. (3. . Massive particles: p2 = M 2 > 0.53) (3. M ǫijk . (3. . p0 > 0. i ǫµνρσ W ρ P σ . Wα ] = [Wµ . which can thus be chosen spacelike. . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. .

3. W i (p) = −W · ni (p)). 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 . 1. .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). (0.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . Together with the four momentum states thus can be labeled as |M.65) . More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). . ms . 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. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3.e. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). p0 > 0. (3.64) such that in an arbitrary frame where the four vectors p. 2 . the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. 0. show that W /M can be interpreted as the intrinsic spin. 0. 0. . n1 (p).61) one sees that S i (p) i =− W2 . 1. |p3 |).57) (i. The commutation relations [W i (p). 0. 0. 0).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. i (3. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. In that case the vectors ni (p) are just the space directions and W = (0. M J) with components W i = (M/2)ǫijk M jk . 1 2 (3.57 The best insight in the meaning of the operators W µ is to sit in the rest frame of the particle and put P µ = (M. M2 (3. 1.e..58) i. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. M2 (3. (0. We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p).59) (3. s. 0. (3. p. (1. Having made a covariant decomposition. 0). W j (p)] = i M ǫijk W k (p).

P µ . We don’t know of any physical relevance. W 2 (p)] [λ(p). But if M = 0 one has a diﬀerent algebra (Eq.Groups and their representations The above does actually include the massive case.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. Note that angular momentum does not contribute to helicity as L · p = 0.70) as a commuting set with zero eigenvalues for P 2 and W 2 .67) (3.1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. implying continuous values for the spin (actually for W i = M S i ). This state is in physical applications nondegenerate and is denoted by |0 .66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). Thus we have P 2 . This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). however. i W 2 (p). |p.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). W 2 (3. The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p).13. W 2 and λ is derived from the general commutation relations for [Wµ . The algebra of W 1 . Exercises Exercise 3. which can take any integer or half-integer value. and gives [W 1 (p). The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. W 2 (p)] = = = i M 2 λ(p). 1 2 24 (3. Wν ].67 has a vanishing right hand side). where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). 3. thus.73) (3. W 1 (p)] [λ(p). −i W 1 (p). U (Λ. A massless particle. 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 .68) (3. W µ (p) = λ(p) P µ . 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). λ . (3. The vacuum: pµ = 0. |p| (3. Thus 2 λ(p) = J ·p . 3. The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with W 0 (p) = − 1 ǫ0ijk Mij pk = J · p. a)|0 = |0 . (3. is characterized by its momentum and the helicity.71) which is called the helicity. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a).

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

(b) In quantum mechanics one has the Ehrenfest relation i dO = [O. derive the commutation relations and show that the algebra is isomorphic to (i. indeed assures d K /dt = 0.e. [ri . Exercise 3. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. and U (u) p U −1 (u) = p − m u. These do not matter in the non-relativistic limit (c → ∞). there is a one to one mapping) the algebra in Eqs 3.67 to 3. they describe r −→ r′ = r − u t while t′ = t. . in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. H] + i dt ∂O . Hint: for the Hamiltonian. (c) Give the unitary operator for boosts. = r × p + s. r2 ) inevitably leads to interaction terms in the boost operators. f (p)] = i ∇p f (p). the others vanish. = p. Exercise 3. This is possible for a single particle.Groups and their representations 26 Exercise 3.e.e. consistent with the (canonical) commutation relations of a quantum theory. Determine the commutation relations for the corresponding operators Ki in quantum mechanics.5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame.69 of the Pauli-Lubanski operators for massless particles. sj ] = [pi . (a) In classical mechanics. if you don’t want to do this in general. rj ] = [pi . sj ] = 0. i. although that in this case is a relation that is found already from the Ehrenfest relation for O = ri . the operator U (u) that gives U (u) r U −1 (u) = r − u t. although the presence of an interaction term V (r1 .7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. Comment: extending this to more particles is a highly non-trivial procedure. that’s why many-particle non-relativistic quantum mechanics is ’easy’. sj ] = iǫijk sk . boost invariance leads to a conserved quantity.. show ﬁrst the operator identity [r. = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. but it can be done.6 Consider rotations and translations in a plane (two-dimensional Euclidean space). ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). you might just check relations involving J or K by taking some (relevant) explicit components. momentum and spin operators e satisfy the canonical commutation relations. From either of these. pj ] = [ri . pj ] = iδ ij and [si . i. p×s 1 (rH + Hr) − t p + . [ri . H P J K = p 2 c 2 + m2 c4 . Do this by showing that the set of operators.

4) (4. (j. (4.6) (4. B j ] = i ǫijk Ak . 2 2 σ σ J = . j ′ ). K = −i .9) (4. C). i ǫijk B k . This tells us how to ﬁnd the representations of the group.7) From the considerations above.1 The Lorentz group and SL(2. Aj ] = [B i . becoming the deﬁning representations of SL(2. Special cases are the following representations: Type I : (j. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . With this choice of parameter-spaces the plus and minus signs are actually relevant. respectively. 2 (4.3) (4. 2 2 (4. They will be labeled by two angular momenta corresponding to the A and B groups. 2 1 (J − i K). K = i J (A = 0). C) Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (hermitean) combinations A = B which satisfy the commutation relations [Ai . They precisely correspond to the two types of representations that we have seen before.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2).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.10) . M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . The matrices in SL(2. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. U (ϕ) H(η) (4. similarly as the homeomorphism between SO(3) and SU (2). C) is the group of complex 2 × 2 matrices with determinant one. The group SL(2. 0.5) = 1 (J + i K). K = +i . B j ] = [Ai .Chapter 4 The Dirac equation 4.1) (4. 0) Type II : (0. It is simply connected and forms the covering group of L↑ . j) K = −i J (B = 0). (4.

With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. ǫ−1 = ǫ† = ǫT = −ǫ).16) = σ µ Λ R aµ . ˜ = σ µ Λ B aµ . H ≡ (H † )−1 = H −1 ).15) (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). −σ). C) or L↑ . (4. however. ˜ (4. σ). σ µ ≡ (1. cannot be connected by a Lorentz transformation belonging to L↑ .e. (4.13) 2M (E + M ) (exercise 4. Eqs 4. This is not true for the two-component spinors in SL(2. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2.18) As already discussed for SU (2).18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. The following relations for + rotations and boosts are useful. thus.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.12) ξ → U ξ. 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. while ±ǫη ∗ transforms as a type-I + spinor. (ǫ∗ = ǫ. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. C) and. H(η) = exp(−η · σ/2). thus. This shows that M ∗ ≃ M . In fact. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1.11) Considering ξ and η as spin states in the rest-frame. if there does not exist a unitary matrix S. C). ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ.19) . η → U η. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. (4. ξ −→ Hξ. namely H(η) = exp(η · σ/2). U σ µ aµ U −1 = σ µ ΛR aµ . C). η → Hη. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . 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. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4.1). (4. The Lorentz transformations Λ(M ) and Λ(M ).14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. Λ(ǫ) ∈ L+ . such that M = SM S −1 . One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . there exists the matrix ǫ = iσ 2 . One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). i. (4. one can use a boost to the frame with momentum ˆ ˆ p.

From the deﬁnition of the representations (0. C). m) = η(0. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0.23) u(p. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . (axial vector). −→ −p (vector). (scalar). Thus M ∗ and M are not equivalent within SL(2. the well-known two-component spinor for a spin 1/2 particle.1.17. The i angular momentum operators J are represented by σ i /2. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). but rather the operators W i (p) should be used. We also have seen that the representations (0. The behavior is the same for classical quantities. Within the Lorentz group. but to L↑ .e. For the spin 1/2 representations of the Poincar´ group including parity we. i.e.13. m) = χm . Taking M (Λ) = H(p). e. (vector). where a = (a0 . and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. that have the correct commutation relations. the aij elements of a tensor will not change sign. but by a transformation belonging ↑ ↑ to the class P− . we obtain for the two components of u which we will refer to as chiral right and chiral left components. they can be connected.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. generators. −a). 4. 1 ) and ( 1 . 0) are inequivalent. −→ −r (vector). A parity transformation changes aµ into aµ . For a particle at rest only angular momentum is important and we can choose ξ(0. Although these operators cannot be used to label the representations. 2 2 (4. This provides the easiest way of seeing ˜ what is happening. a) and ˜ a = (a0 . 4. i. they cannot be connected by a unitary 2 2 transformation.g.The Dirac equation 29 where The matrix I2 does not belong to L↑ . etc. ) −→ ( . the boost in Eq. −→ M (Λ) (4. 1 ) and ( 1 . 0).22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. m) = χm uL H(p) 0 . + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 .2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. It has the same eﬀect as lowering the indices. as we have seen in section 4. ǫµνρσ will change sign.20) 4. we have seen that in the rest frame W i (p)/M → J i . 0 χm uR H(p) . therefore. must combine the repe resentations. (pseudoscalar). The representing member of the class P− is the parity or space inversion operator Is . (4. 2 ) and ( 2 . 0) of SL(2. = (4. C) are suitable for representing spin 1/2 particles.

29) which is an explicit realization of the (momentum space) Dirac equation.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p).30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.33) deﬁne for a four vector a the contraction a = aµ γµ / . The ﬁrst indeed is solved. From this one obtains the Cliﬀord algebra for the Dirac matrices. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.27 is known as the Weyl representation. 0 H 2 (p) uR uR . Applying (iγ µ ∂µ + M ) from the left gives (4. {γ µ . 2 (4. The explicit realization appearing in Eq. We will discuss another explicit realization of this algebra in the next section. = (4. (4. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). γ } ∂µ ∂ν + M 2 ψ(x) = 0. (iγ µ ∂µ − M ) ψ(x) = 0.7. Since ∂µ ∂ν is symmetric. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). this can be rewritten 1 µ ν {γ .The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . The general requirements for the γ matrices are thus easily obtained. (4. p (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . which in general is a linear equation in pµ . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. = H −1 (p). As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function 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. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. .28) which is a covariant (linear) ﬁrst order diﬀerential equation. It is of a form that we also played with in Exercise 2. (4. (4.32) suppressing on the RHS the identity matrix in Dirac space. γ ν } = 2 g µν .24) (4. (4.

(4.g. E = +M (twice) and E = −M (twice). Relying on the Pauli exclusion principle for spin 1/2 particles.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 . but with properties such that it can be annihilated by the particle (e. . remains. The Dirac sea forms the vacuum. 4. The second problem encountered before. This can most easily be seen in the particle rest frame.1: The Dirac sea of negative energy states and antiparticles. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). This hole forms an antiparticle with the same mass. 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.35) is indeed positive and j 0 can serve as a probability density.3 General representations of γ matrices and Dirac spinors {γ µ .The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. 4. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ. . γ ν } = 2 g µν . γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. opposite charge). Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. (4.27). (4. one sees that there are two positive and two negative eigenvalues. From the vacuum a particle can be removed.34) for the adjoint spinor ψ ≡ ψ † γ0 .37) Using the explicit form of γ in the Weyl representation (see Eq. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds. the one of the negative energy states. This problem was solved by Dirac through the introduction of a negative energy sea.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices.36) (4. The density j 0 = ψγ 0 ψ = ψ † ψ (4.

while L−1 γ µ L = Λµ γ ν . We note that ψ → Lψ and ψ → ψ L−1 .40) (4. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.41) (4. =ρ ⊗1= 0 −1 3 (4. (γ5 )S.47) (4. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations.3) .46) (4. (4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . γ σ ].42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. the rotation and boost generators in Weyl representation in Eqs 4. γ 2 ]. 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 . −→ Sγµ S −1 . which in the case of the Weyl representation are just the upper and lower components.9 and 4.R.R. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. γ µ } = 0 and explicitly one has 0 1 1 . is 1 1 1 (4. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k .45) (γ5 )W. that project out chiral right/left states. For instance in the Weyl representation (but valid generally). −→ Sψ.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ .48) are projection operators. γ 0 ].R. S −1 . (γµ )S. 0 (4.R. 2 (4.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. = S(γµ )W.43) S=√ . For example. 4! (4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 .

(C)S. Starting with (i/ − M ) ψ(x) = 0 we note ∂ that by hermitean conjugating and transposing the Dirac equation one obtains iγ µT ∂µ + M ψ (x) = 0.6). (C)W. ∂ with the solution ψ c (x) = ηc Cψ (x).55) (4. e.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. x) → x = (t. is a left-handed spinor.R.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. The latter implies that the conjugate of a right-handed spinor. Explicitly. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η .58) where ηc is an arbitrary (unobservable) phase. reverse the charge of the particle. ∂ (4. such that T Cγµ C −1 = −γµ . In S. γ5 ] = 0. x = (t.53) (4. T (4. parity. This symmetry does not change the spin 1/2 nature.g. ψ= (4. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . but it does. ψR .g. = = 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. One has ψ c = −ψ T C −1 . C is real and one has C −1 = C † = C T = −C c and [C.56) (i/ − M ) ψ c (x) = 0.R. which is again a solution of the Dirac equation. Some properties of C are C −1 = C † and C T = −C. −iσ 2 0 −ǫ = . for instance.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. called charge conjugation.52) η ξ one can apply space inversion.59) −→ ψ c = C ψ = η ǫ ξ∗ .R.54) e. that brings ψ → ψ c . −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. (i/ − M ) ψ(x) = 0. usually to be taken unity. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. As with parity we look for a transformation. (or W. ψ= (4. In any representation a matrix C exist.) and all representations connected via a real (up to an overall phase) matrix S.R. T (4. 0 −ǫ 0 0 −ǫ 0 .57) (4. = 0 ǫ iσ 2 (4. ∂ (4.

The ¯ ¯ solutions are normalized to u(p. ⇔ (/ − M ) u(p) = 0. s′ ) = −2M δss′ . The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. s) ei k·x d∗ (k.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. s ) = v (p. s)u(p.63) u(p. s) and C v T (p. in which γ 0 is diagonal (see discussion of negative energy states in section 4. the spinors satisfy C uT (p. ∂ −iσ · ∇ −i ∂t − M (4. (4. (2π)3 2Ek (4. s)¯(p. be obtained by a Lorentz transformation. ψ(x) = v ±s (p) ei p·x .70) = Ps = ˆ 0 1∓σ·s 2 2 .. the best representation to describe particles at rest is the standard representation. 2M 2M (4. s) = v(p. with E = Ep = + p2 + M 2 . s)u(p. ψ(x) = u±s (p) e−i p·x . where u satisﬁes −σ · p Ep − M u(p) = 0. In that frame is a spacelike unit vector sµ = (0. s) u = . s). ¯ (4. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. ⇔ (/ + M ) v(p) = 0. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s) . s) = u(p. s)¯(p. In an arbitrary frame one has s · p = 0 and s(p) can e.64) (4. s′ ) = 2M δss′ . s) v = .The Dirac equation 34 4. v (p.62) where χs are two independent (s = ±) two-component spinors. s ) = v † (p. s ) = 0. s)u(p.68) (4. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors.1). s). s)v(p. ¯ ′ ′ Explicit solutions in the standard representation are χs . where v satisﬁes σ·p −(Ep + M ) v(p) = 0. After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) = s d3 k u(k.65) (4. ¯ ¯ † ′ u (p. 2M 2M −/ + M p v(p. s) = Ep + M σ ·p Ep +M χs . s)v(p.69) = − s In order to project out spin states.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. s)v(p. v(p.61) σ·p −(Ep + M ) There are also two negative energy solutions. s) = u(p.60) ψ(x) = 0.g.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s′ ) = 2Ep δss′ . s) + v(k. the spin polarization vector in the rest frame is the starting ˆ point. s) e−i k·x b(k. p (4.

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

we will give one example. s′ )(γ µ )ij uj (k.) (2) Tr(1) = 4. In principle one can take a representation and just calculate the quantity u(k. s′ )) .s′ ∗ (4. s)γ µ u(k ′ . s)¯k (k. namely the calculation of the quantity 1 ∗ (¯(k. s)γ ν u(k ′ . k k 2 . however. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν .5 γ gymnastics and applications An overview of properties can be found in many text books. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ .s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . etc.) In order to show the use of the above relations. k ′ ) = 2 ′ s. more convenient ¯ to use the projection operators introduced earlier. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ . (6) Tr(γ5 ) = 0. a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b. s′ ) ¯ u s. s′ )¯i (k ′ . (Use (γ5 )2 = 1 and pull one γ5 through. s)¯k (k. (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) .g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . s). s′ ). u u (4. 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′ )) (¯(k. s)γ µ u(k ′ . s′ )) = u(k ′ . u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ .76) ul (k ′ . (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b.75) Lµν (k.The Dirac equation 36 4. It is. (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. s)γ µ u(k ′ . For this we ﬁrst have to prove (do this) that (¯(k.g µρ g νσ + g µσ g νρ ). s′ )(γ µ )ij uj (k. s)(γ ν )kl u u s. s)(γ ν )kl ul (k ′ . (5) γ µ γ λ γ ν γ ρ γ σ γµ = 2 γ σ γ λ γ ν γ ρ + γ ρ γ ν γ λ γ σ . (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2.s which appears in quantum electrodynamics calculations (see Chapter 10) for the emission of a photon from an electron changing its momentum from k to k ′ (k 2 = k ′2 = m2 ). s′ )γ µ u(k. (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. (10) γ µ γ ν = g µν − i σ µν (11) γ µ γ ν γ ρ = S µνρσ γσ + i ǫµνρσ γσ γ5 with S µνρσ = (g µν g ρσ . (8) σµν ≡ (i/2)[γµ . (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ).

γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.4 Show that in P+ ≡ s=±1/2 u(p.13.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. 4 2 Exercise 4.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. γ σ ] = σ ρσ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. Do this e.1 Prove Eq.6 Apply space-inversion. 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 . H(p) = exp η·σ 2 = M +E+σ·p . x ˜ . x → x.g. s) and v(p. ˜ Exercise 4. (4. by letting them act on the four basis spinors u(p. Excercise 4. 4. Show that the equality holds.77) Exercises Excercise 4. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ .g. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. s). Exercise 4. s)¯(p. 1 i −i S ρσ = − [γ ρ . 0 if n is odd. s) u p /+M = 2M 2M both sides are projection operators.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . a 4 a · b. −x) is also a solution of the Dirac equation. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). i S ρσ ] = g ρµ γ σ − g σµ γ ρ . e.5 Show starting from the relation {γµ . where x = (t. of which only the traces of four and two gamma-matrices are nonzero.

there is no current ﬂowing through the surface of the cavity. Do you understand why? The calculation of the life-time is completed in chapter 11. you need to determine the corresponding conﬁnement condition for ψ. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. is a stationary solution of the (free) Dirac equation. In this solution the wave number k = |k| √ = E 2 − M 2 and χm a two-component spinor while j0 and j1 are spherical Bessel functions. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity.The Dirac equation 38 Exercise 4. Note: To prove this.511 MeV).57 MeV. The pion has spin 0. . which is the most important factor needed to ﬁnd the decay width (inverse lifetime). i. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. For such a cavity nµ = (0. We will not pursue this chiral freedom at this point and take α = 0. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. ν p Note: how such parametrizations work will be explained in chapter 8.5 and M → ∞. (b) Next.66 MeV. Calculate for now just |M |2 . Calculate the value of k for M R = 0. r). 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. M R = 1.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0.e.8 × 10−6 . me = 0. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). the decay would even be forbidden. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . we want to conﬁne this solution to a spherical cavity with radius R. mµ = 105. Our starting point is the transition amplitude given by M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ).7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. namely i / ψ = eiαγ5 ψ. n It turns out that there is actually a whole family of conditions that provide conﬁnement. Excercise 4.

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. ±(1) are obtained by boosting the restframe vectors.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. λ) .2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. In an arbitrary frame ǫµ (k) with λ = 0.Chapter 5 Vector ﬁelds and Maxwell equations 5. µ (2π)3 2E µ (5.± 5. namely ǫµ (k0 ) = (0. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . λ) + ǫ(λ)∗ (k) eik·x c∗ (k. (2 + M 2 )Vµ = 0. The vectorﬁeld is therefore (λ) suited to describe spin 1. (5.7) . In the restframe. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0.1 Fields for spin 1 In section 3.1) The Lorentz condition implies that k µ ǫµ (k) = 0.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. there are then indeed three independent possibilities.6) where Aµ is the four-vector potential (5. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . ǫλ ). just as the two-component Dirac spinors did for spin 1/2. M2 (5. A). These can describe spin 1 states in the rest-frame. 0). with ∂µ V µ = 0 (Lorentz condition).5) −E B2 .4) λ=0. 39 (5. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. These are solutions of the Klein-Gordon equation with in addition a condition. where k = k0 = (M.2) (5.

j) the Maxwell equations read ∂µ F µν = j ν . As remarked above. for M = 0.10) the electric and magnetic ﬁelds are unchanged. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . In that case one also has ∇·A = 0 or ki ǫi (k) = 0. if in Eq. moree over. One possible choice is the gauge A0 = 0 (5.e. ˜ x Note that via parity transformation one obtains another solution.16) (5. −i. Under a gauge transformation Aµ −→ Aµ + ∂µ χ.13) Hence. 0. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. (5. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. ∂µ Aµ = 0. it will vanish at all times. ∓1. It states the absence of magnetic charges and Faraday’s law. Furthermore. 2Aµ = 0. i. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ .9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. Taking the divergence of this equiation. |k|). (5.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. 5. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). 0.11) This equation is not aﬀected by a gauge transformation. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0.15) (5.17) (radiation. if ∂µ Aµ = 0 for large times |t|. . there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0).Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. in accordance with what we found in section 3. Although the Lorenz condition is a constraint.15 the d’Alembertian is replaced by 2 + M 2 . 40 (5. (5. (5. transverse or Coulomb gauge). AP (x) = Aµ (˜). 0). The equation in vacuum. Therefore if k µ = (|k|. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0.12) (5. For electrodynamics one has furthermore the freedom of gauge transformations.5.14) of which the solutions give the Li´nard-Wiechert potentials. (5. only two independent vectors remain ǫ(±) µ (k) = (0. in which case one has 2Aµ = jµ .8) (5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld.

1: The Aharonov-Bohm experiment 5. F = e E + ev × B (5. cos ϕ. while the electromagnetic ﬁeld Aµ is not. The situation.19) are gauge invariant. This phase diﬀerence is actually observed. 2 2 2 BR2 BR2 y BR2 x A= . may look a bit akward.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. nevertheless. To be precise.1. 5. Inside solenoid: Outside solenoid: B = B z. It is however. . ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. but A = 0 (hence there is structure in the vacuum). sin ϕ.0 . nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . This causes an interference pattern as a function of ∆x. 0). The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). where Aµ = 0. but where E = B = 0. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. This occurs in the region outside the solenoid.Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. (5. = exp −i The phase diﬀerence between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. 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 . ˆ B = 0.3 The electromagnetic ﬁeld and topology∗ The electric and magnetic ﬁelds E and B as appearing for instance in the Lorentz force on a moving charge. 0) and ϕ = (− sin ϕ.0 . . λ L − λ λ or ∆x = (L− λ/d)δ. . A= By Bx Br ϕ= − ˆ . In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). illustrated in ﬁg.

2: The topology of the space in the Aharonov-Bohm experiment. Therefore the gauge transformation must be uniquely deﬁned in the space one is working in. If this would be be happening in empty space one would be in trouble. Consider a superconductor. the number of times they go around the defect) cannot be continuously deformed into one another. i. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. however. 1 The occurrence of nontrivial possibilities. implying that it must be single-valued1 . forming a simple (connected) space. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs .e. it can be obtained from the situation A = 0 by a gauge transformation. A = ∇χ with χ= BR2 ϕ 2 (5. Another situation in which the topology of space can be used is in the case of superconductors.21.21) (ϕ being the azimuthal angle). minimizing the space occupied by B ﬁelds. Now back to the Aharonov-Bohm experiment. The only way for ﬂux tubes to be formed and move around without global consequences is when each ﬂux tube contains a ﬂux Φ such that ∆δ = (e/ )Φ = n · 2π.e.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. nonobservable phases ϕ = 2π n. The AharonovBohm experiment shows a realization of this possibility. Therefore space outside the solenoid doesn’t care that χ is multi-valued. giving no observable phase2 . By going around an arbitrary loop a diﬀerent number of times the electron wave function would acquire diﬀerent phases or it would acquire an arbitrary phase in a point by contracting a loop continuously into that point. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. 5. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . 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. organizing itself in tiny ﬂux tubes (Abrikosov strings). ϕ=2π The function χ in Eq. In such a space loops with diﬀerent winding number (i. This situation that A can be written as ∇χ is always true when ∇ × A = 0. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually.

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

The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current.1) and as an example the lagrangian L(x.6) The ﬁrst term leads to the Euler-Lagrange equations. recall classical mechanics with the action t2 S= t1 dt L(x. ∂x (6. ∂x ˙ (6. These equations can be obtained from a lagrangian using the action principle.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 . As an example.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. x) = K − V = ˙ 1 mx2 − V (x).1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). (6.Chapter 6 Classical lagrangian ﬁeld theory 6. d dt ∂L ∂x ˙ = ∂L . ˙ p= ∂L . ˙ 2 (6. t1 (6. thus t′ = t + δτ.5) with ∆x(t) = δx(t) + x(t) δτ . x) ˙ (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. and the total change x′ (t′ ) = x(t) + ∆x(t). x (t) = x(t) + δx(t).8) 44 .3) (6.

13) (6. . t1 (6.2. (R3 .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). but which in 3 dimensions includes conserved quantities related to rotations and boosts. + ∂L ∆x − H δτ ∂x ˙ t2 . t2 ). which in classical mechanics vanishes at the boundary) does not play a role. (6. 6. ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . − d3 x . In general any continuous transformation. = ∂R (R3 .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ .6 can be rewritten as δS = . . . indicated with δF [φ]/δφ should pose no problems.e.9) which is done because the ﬁrst term (multiplying ∆x.12) Here R indicates a space-time volume bounded by (R3 . think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). . H is a conserved quantity. We will come back to it in a bit more formal way in section 9. Consider a lagrangian density L which depends on these functions. ˙ x The second term in Eq. dσµ . .14) (6. R (6. ∂µ φ(x)). with ∆φ(x) = δφ(x) + (∂µ φ)δx .Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. .t2 ) d3 x . ∂µ φ(x). . ˙ (6. i. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x).18) 1 Taking a functional derivative. 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 . x) ≡ p x − L . L (φ(x). their derivatives and possibly on the position. The hamiltonian H is deﬁned by H(p. ∂µ φ(x)) = R d4 x L (φ(x). = φ(x) + δφ(x) (6. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ .t1 ) (6. ∂µ φ′ . . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds.15) d4 x′ L (φ′ . λ=0 (6.11) λ=0 of which p and H are the simplest examples related to space and time translations. . The resulting variation of the action is δS = R µ ′ = xµ + δxµ .17) Furthermore the variations δφ and δ∂µ φ contribute to δS. t1 ) and R3 . δ(∂µ φ) (6. x′ ) − d4 x L (φ. x). ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. . ∂µ φ. Variations in the action can come from the coordinates or the ﬁelds.

Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.20) (6. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. (6.19) δ(∂µ φ) δφ 6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. (6.21) which diﬀer only by surface terms. δL δL ∂µ = . 2 ψ = 0. the second term is irrelevant.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. 2 ∗ (6.2 Lagrangians for spin 0. The integrand of the ﬁrst term must vanish.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration).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. ∂ (6. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). One easily obtains (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. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. (2 + M )φ (x) = 0.27) (6.23) (6. (6. leading to the Euler-Lagrange equations.29) . the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. 2 (6.25) (6. Using the variations in ψ (in the symmetric form).24) (6. leading to (2 + M 2 )φ(x) = 0.

and lefthanded ones. Peshkin and Schroeder or Exercise 12.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). 2 (6.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. 2 αβ = −βα.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. or equivalently separate the ﬁeld into right. since a lagrangian with both Dirac and Majorana mass terms can be rewritten as the sum of two Majorana lagrangians after redeﬁning the ﬁelds (See e.g. The reasons for Grassmann variables will become clear in the next chapter. The Majorana case is in fact more general. (6.36) (6. Using the twospinors ξ and η.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . which comes because we in essence use ’real’ spinors. and thus (β ∗ α)∗ = α∗ β = −βα∗ .39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . (6.31 is given by LM (Dirac) = −M ξ † η + η † ξ .38) (6. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR .34) (6.7). but note the factor 1/2 as compared to the Dirac lagrangian. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . 1 M η † ǫη ∗ − M ∗ η T ǫη . ∂ ∂ 2 2 (6. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR .32) There exists another possibility to write down a mass term with only one kind of ﬁelds.g. (6.31) showing e. it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left ﬁelds ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors. namely2 LM (Majorana) = + (6. the mass term in the Dirac lagrangian 6. We note that of which the left part coincides . ψL ≡ (ψL )c = C ψL = 0 (6. Vector ﬁeld From the lagrangian density L 1 1 λ 2 = − Fµν F µν + M 2 Vµ V µ − (∂µ V µ ) 4 2 2 1 µ = V (∂ 2 + M 2 )gµν − (1 − λ)∂µ ∂ν V ν .

the presence of a symmetry that leaves the lagrangian invariant.44) δL ∆φ + Θµν (x)δxν . discussed in the next chapter. In this case. t). which is the space-integral over the zero-component of a conserved current. δ(∂µ φ) For continuous transformations. Q= d3 x J 0 (x. Q(t1 ) = Q(t2 ).+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν .42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . (6. (6. .47) .18. . these conserved quantities become operators. 6. These do not aﬀect the dynamics and will give rise to conserved currents. As an example for the classical ﬁeld.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . Returning to δS the surface term is rewritten to δS = R d4 x .3 Symmetries and conserved (Noether) currents Next. In particular when the lagrangian is invariant under symmetries. + ∂R dσµ dσµ ∂R ≡ where d x . δ(∂µ φ) (6.41) (6. 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. (6.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .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 .43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (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). δ(∂µ φ) (6. we consider the second term in Eq.. which in a consistent picture precisely generate the symmetries. one can consider the variations at the initial or ﬁnal surface. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. (6. requires the presence of a conserved quantity. t).σ1 with σ = (R3 . (6. ∂µ J µ (x) = 0. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0.

Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld. (6.49) These currents are conserved as discussed already in chapter 2.53) These are the energy and momentum. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. we obtain (since δxµ = 0). t) is the conserved charge (Q = 0). ˙ Q = d3 x j 0 (x. In the next section we will see the quantity show up as the charge operator.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). . 1 (6. (6.54) (6. From Noether’s theorem one sees that the current Θµν ǫν is conserved. H = d3 x Θ00 (x). (6. The integral over the zero-component. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . (6. 6.g.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. ↔ (6. 2 (6. = 0. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν .55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). including e the space-time translations and the Lorentz transformations.50) (6.52) The behavior of the ﬁeld under translations (δφ) is governed by the translational behavior of the argument in such a way that ∆φ = 0. 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. Summarizing. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). e.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν .3) j µ = i e φ∗ ∂µ φ.51) (6.

• Vector ﬁeld Aµ : −i Sρσ = aρσ .69) i eΛ(x) φ (x).g.3). (6.65) M µρσ = T µρ xσ − T µσ xρ . ∂µ T µν (6. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). (6.e. 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 . −i eΛ(x) ∗ i eΛ(x) (6.64) (6. independent of x.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . (6. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. In general this tensor is not symmetric.6) T µν = T νµ . the angle of the transformation depends on the space-time point x. however.61) A ﬁnal note concernes the symmetry properties of Θµν . The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).67) i. 2 (6. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. Deﬁning T µν = Θµν − ∂ρ Gρµν . in which the U (1) transformation involves an angle e Λ.58) (6. is not invariant under local gauge transformations. (6. φ(x).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. e.70) .66) In section 6. ∂µ φ(x) + i e ∂µ Λ(x) e (6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. 6.63) µν = ∂µ Θ . where H ρµν is the quantity appearing in M ρµν .3 we have seen global transformations or gauge transformations of the ﬁrst kind.68) (6.59) (6. one has a tensor that satisﬁes (exercise 6. Any lagrangian containing derivatives.

∂µ φ) =⇒ L (φ.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 . D ∂ A (6. 1 L (φ.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 . For this purpose it is necessary to introduce a vector ﬁeld Aµ . i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. Dµ φ) − Fµν F µν . 4 (6. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x). 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. −e j µ Aµ = −e ρφ + e j · A. 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 .Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).72) (6. ∂ 2 4 (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. 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. (6.73) (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. (6.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.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).78) . (6. As an example consider the Dirac lagrangian.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.

∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). Exercise 6.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . where j µ = e ψγ µ ψ. ∂µ F µν = j ν .Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . Exercise 6.e. −eψγ ν ψ.4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. . φ → eiΛ φ.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 . ∂µ F µν = j ν . 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. ↔ 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. and deduce what is the charge of the antiparticle as compared to a particle. give the equation which is satisﬁed by ψ c = Cψ . i. giving the Maxwell equation coupling to the electromagnetic current. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved.79) Exercises Exercise 6. (6.

t) = Enr ψu (r. Hint ≡ −gs µe · B = −gs Qe s · B. . use ∂µ T µν = 0 and ∂µ M µρσ = 0. This relation is known as the Pauli equation. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. 2M 2M with Enr ≡ E − M ≪ M . A) (see also exercise 6.4) for a stationary solution in the non-relativistic (|Enr | ≪ M ) and weak-ﬁeld (|A0 | ≪ M ) limits yields in the standard representation for the ’upper (two) components’ ψu . 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. 2me Exercise 6.6 (optional) Prove the properties of the tensor T µν in section 6. Comment: 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 . t).Classical lagrangian ﬁeld theory 53 Exercise 6. in order to show that Θρσ − Θσρ = ∂µ H µρσ . ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ .4.

6. Further. (7. ˙ 2 (7. however.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. 1 qx (t) = 3 d3 x φ(x. The bilinear product for the conserved quantities. p]P . labeled by the position.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t). p]P = 1.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q).5) ∆ x ∆3 x 54 . q) = p q − L(q. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. dq dp dq dp (7. q) = ˙ 1 mq 2 − V (q). is the Poisson bracket [A. p] = i.11) found through Noether’s theorem. [q. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. q(q.Chapter 7 Quantization of ﬁelds 7.2) H(p. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. in our example p = mq. q) also considered in the previous chapter. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. B]P = dB dA dA dB − . Quantization of the theory is achieved by imposing canonical commutation relations between q and p. namely [q.3) For variations δO(p. there are conserved ˙ ˙ quantities Q (Eq. Q]P . (7. dλ Examples are dO = [O. H]P dt and dO = [O. t). starting from the lagrangian L(q. An immediate generalization for ﬁelds can be obtained by considering them as coordinates. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. (7. ˙ L(q.

The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). φ(y. Sometimes it is useful to keep in mind that. t).Quantization of ﬁelds 55 etc. 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.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. ∂µ φ(x)) = ˙ d3 x L (φ.11) (7. a) = Rj (Λ−1 ) φj (Λ−1 x − a). the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. • Poincar´ invariance e The full variation of the ﬁelds.14) In fact. Π(y. for the scalar ﬁeld theory. t). ∇φ) (7. albeit with ’sharp’ functions. φ. t). 2 2 2 (7. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).8) (7. formally. t)] = i δ 3 (x − y). • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). i U † (Λ.2.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.16) . the discretization procedure above is an explicit example. t)] = [Π(x.12) (7.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . ∆φ. Φ(f ) = d4 x φ(x) f (x). ˙ δ φ(x) (7. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . (7.7) d3 x L (φ(x).15) with furthermore the relations [φ(x. t)] = 0. qx . ∂ qx ˙ ∂ qx ˙ (7. As an example. Π(y. for symmetry transformations. qx+∆x ) . a) φi (x)U (Λ. The essential conditions to consistently quantize a theory then are the following. discussed in the section 6. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. (7. (7.6) ∆3 x L x (qx .

2 2 (7. [N.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† ] = 0. ω a† a + 2 2 ω i − [Q. P ] = i.23) The hamiltonian in this case can be expressed in the number operator N = a† a.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . Φ(g)] = 0. a† ] = a† .24) It is straightforward to ﬁnd the commutation relations between N and a and a† . φ(y)] = 0 if (x − y)2 < 0.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. and [N. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . (7.17) (Sµν )i φj (x). [φ(x). [a. Pµ ] = [φ (x).20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. sometimes referred to as second quantization. [Q. j 2 one has [φi (x). 2 i ωµν (S µν )i . Mµν ] = i i∂µ φi (x). a† ] = 1.Quantization of ﬁelds 56 or if U (Λ. a] = −a. Q] = [P. i(xµ ∂ν − xν ∂µ )φ (x) + i (7.25) .4. 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. 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). j (7. a] = [a† .18) with S µν given in 6. The measurements cannot inﬂuence each other or [Φ(f ).19) 7. P ] = 0 (7. (7. P ] 2 (7. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. (7. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν .

φ(x )]x0 =x′0 = [Π(x). but one needs to realize that in that case k 0 = Ek = k2 + M 2 .31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). the real scalar ﬁeld φ(x) becomes d3 k (7. which represents the ’number of particles’ with momentum k.e.32) (7. Π(x )]x0 =x′0 = 0. (2π)3 2Ek (7. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. and we see that a state |0 must exist for which N |0 = a|0 = 0.30) [φ(x). N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . a(k ′ )] = [a† (k). a† (k ′ )] = 0. Note that we will often write a(k) or a† (k).like’ commutation relations between a(k) and a† (k ′ ).33) . (7. is equivalent with [a(k). and (7. † √ a and a act as raising and lowering operators.26) a(k) e−i k·x + a† (k) ei k·x . (n − 1) a|n † i. e. i.e. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . 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 . φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .28) (7.g. 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. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. 7.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. [φ(x). Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).

44) p|φ− (x)|0 = ei p·x . . . The rest of the states are then obtained from the groundstate via the creation operator.34) where V = (2π)3 δ 3 (0) is the space-volume.35) where the vacuum contribution disappears because of rotational symmetry. (2π)3 2Ek (7.46) (7.37) (7. The 1-particle wave function e−ip·x is obtained via 0|φ(x)|p p|φ(x)|0 = = 0|φ+ (x)|p = d3 k 0|a(k)a† (p)|0 e−i k·x = e−i p·x (2π)3 2Ek (7. 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). (2π)3 2Ek (7. d3 k |k k|.40) This procedure is known as normal ordering.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. a(k)|0 = 0.42) (7. d3 k a(k) e−i k·x .39) The problem with the vacuum or zero-point energy.45) (7. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. (2π)3 2Ek d3 k a† (k) ei k·x . (2π)3 2Ek (7. and multiparticle states |(k1 )n1 (k2 )n2 . |0 . n1 ! n2 ! (7.43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y).e. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . (2π)3 2Ek (7.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7.41) (7. (2π)3 (7. deﬁning particle states |k = |k (with positive energy. This term will be adressed below. . which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). . = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . k 0 = Ek = k2 + M 2 ). writing all annihilation operators to the right of the creation operators. 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). . |k = a† (k)|0 .36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). is solved by subtracting it as an (inﬁnite) constant. which now contains an inﬁnite number of oscillators. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. i.

= − (2π)3 2Ek (7. x) = 0 and hence ∆(x) = 0 for x2 < 0.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). (2π)3 2Ek ≡ i∆− (x − y). i. (7. φ+ (y)] d3 k e−i k·(x−y) . ∆(x) = − ǫ(x0 ) δ(x2 ).50) or as integrals over d4 k.18).53) t=0 .Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (2π)4 (7. (iii) ∆(0.51) (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.49) (7.48) (7. d3 k = (2π)3 2Ek = (7.17) and (7. φ− (y)] ≡ i∆+ (x − y). 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 ). The last item to be checked for the scalar ﬁeld are the causality condition. d3 k ei k·(x−y) = −i∆+ (y − x). In order to calculate [φ(x). (7. 2π (7.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .52) The result for the invariant commutator function is [φ(x). φ(y)] consider µν [φ+ (x). i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.e.56) = −δ 3 (x). that they satisfy the required commutation relations in Eqs (7. (2π)4 d4 k − θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7.

66) . (2π)3 2Ek (7. (2π)3 2Ek (7.63) i. = ∂{µ φ∗ ∂ν} φ − L gµν . 2 From the lagrangian density (7. ↔ (7.57) (7.59) (7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld.e.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). 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) . which is equivalent to the relations (only nonzero ones) [a(k). extending it with the charge operator and the introduction of particle and antiparticle operators.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ. The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x .58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) . b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). a† (k ′ )] = [b(k). (2π)3 2Ek (7.Quantization of ﬁelds 60 7.60) = i φ∗ ∂µ φ. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . we will consider it as a repetition of the quantization procedure. φ† (y)] = i∆(x − y).65) 7.62) (7.61) (7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). (7.

(7. s) ei k·(x−y) x0 =y 0 . s)u(k. .69) (7. (7. jµ Θµν = = ψγµ ψ. (2π)3 2Ek (7. d† (k ′ .Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. s)¯(k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . s)ei k·x .76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. s)e−i k·x .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) : .74) (7.67) i ψ / ψ − M ψψ gµν . s)ei k·x + d(k. s) e−i k·(x−y) j s † + vi (k. The solution is the introduction of anticommutation relations.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). s)v(k.e. s). s)e−i k·x + d† (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) . ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. i.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. † {ψi (x). b† (k ′ . s). interchanging creation and annihilation operators. {b(k.73) (7.70) where the last line is obtained by using the Dirac equation. u v (2π)3 2Ek (7. s)¯(k. s)vj (k. s)u† (k. (2π)3 2Ek d3 k b† (k. s′ )} = {d(k.78) (7. ↔ i ψγµ ∂ν ψ − 2 (7.71) (7.77) Note that achieving normal ordering. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. The quantized ﬁelds are written ψ(x) = s d3 k b(k.75) (7. (2π)3 2Ek (7. ∂ 2 ↔ (7.

(7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.80) 3 = δ 3 (x − y) δij . 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).85) which reﬂects the gauge freedom.83) From the lagrangian density the canonical momenta are = = (7. ∂ (7.82) where i∆1 = i∆+ −i∆− .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7. ∂ (7. For the scalar combination {ψi (x). It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . ψ j (y)} at arbitrary times one has {ψi (x). 4 Π0 Πi δL = 0. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. we would have obtained [ψi (x).86) (7. but has the problem of being noncovariant.6 The electromagnetic ﬁeld 1 L = − Fµν F µν . ψ j (y)] = (i/x + M )ij i∆1 (x − y).84) (7. 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). 4 2 This gives the equations of motion discussed before. (7. ˙ δ Ai (7.88) . one has † {ψi (x). −λ(∂ρ Aρ ). as the vanishing of Π0 induces a constraint.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. ˙ δ A0 δL = F i0 = E i .

95) Exercises Exercise 7. (7. i.94) Choosing k µ = (|k 3 |. for which it is suﬃcient that ∂µ Aµ |A = 0. 7. c† (k ′ .90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). containing a time-like photon. It gives 3 k µ ǫ(λ) (k)c(k. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. 6. Πν (y)]x0 =y0 = i gµν δ 3 (x − y).89) with four independent vectors ǫµ . where Q is the conserved quantity in Eq.91) (7. µ λ=0 (7. The canonical equal time commutation relations are [Aµ (x).1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq.e. Πν (y)] = 0.e. λ)|A = 0. 0. λ) − c† (k. λ). λ)ei k·x . 0) . µ λ=0 (7. i. ˙ ˙ .Quantization of ﬁelds 63 for physical states |A and |B . 0.3) satisfy the conditions of a Lie algebra. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). rj ]P .93) which does exhibit problems with the time-like photon. (b) Proof for a quantity O(p. and are equivalent with [c(k. (7. which reproduce the Hamilton equations. and [ℓi . a longitudinal photon and two transverse photons. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). These problems are solved by the Lorentz constraint between physical states given above. 0)c(k. (d) Use Poisson brackets to calculate q and p. pj ]P . ′ (7. pj ]P . In fact the commutation relations imply ˙ [Aµ (x). one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.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. j = 1. Aν (y)] = [Πµ (x). q) that dO/dλ = [O. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k.11. [ℓi . λ)c(k. what are [ri . Q]P . The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. (7. 2. λ)e−i k·x + 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. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ.

Pµ ] = i∂µ φ(x). (b) Check the above commutation relation [φ(x). (e) Show that ∆R (x) = 0 if x0 < 0.4 (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. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). .3 (a) Show that the correct implimentation of symmetries in the Hilbert space. (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−i a requires [φ(x). 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. we can write for ∆R an integral where the k 0 integration remains along the real axis.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.Quantization of ﬁelds 64 Exercise 7. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration). Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. 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. Give also the expression for ∆C . (2 + M 2 ) ∆inhom (x) = −δ 4 (x). ∆hom (Λx) = ∆hom (x) if f is an invariant function. (2 + M 2 ) ∆hom (x) = 0.

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

2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. 8. starting with the Dirac equation for ψ(x). writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). T. −r).1 Parity xµ = (t. parity (P). time reversal (T) and charge conjugation (C). We can determine A. and C quantity t r xµ E p pµ L s ˆ λ = s·p P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ T -t r −˜µ x E -p pµ ˜ -L -s λ C t r xµ E p pµ L s λ 66 . x (8. ˜ (8.1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x). The consequences of P. T and C for classical quantities is shown in the table 1. Both ψ p and ψ satisfy the Dirac equation. Table 8. (iγ µ ∂µ − M ) ψ(x) = 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.

7) = λ d k x x ηP b(k.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). λ) e−ik·x − d† (k.13) ψ(x) −→ ψ c (x) = ηC C ψ (x).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. λ)v(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. m).9) (8.11) (8. (2π)3 2Ek (8. the above operation reverses the sign of λ).6) (check this for the standard representation. . −λ). m). λ) P −1 = −η ∗ d(k.4) (8.3) (8. x The latter behavior of the vector ﬁeld will be discussed further below.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. m) = = ˜ u(k. λ) e−ik·˜ + d† (k. λ)γ0 v(k. 8. λ) Pop = ηP b(k. In the same way as the Fermion ﬁeld.e. if helicity λ is used instead of the z-component of the spin m. λ)γ0 u(k. iγ µ† ∂µ − M ψ(˜) = 0. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. λ)u(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. λ)u(k. ˜ −v(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. −λ) e−ik·x − d† (k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). x (8. λ) eik·x . one can also consider the scalar ﬁeld and vector ﬁelds. ˜ Pop d(k.10) i. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. m) γ0 v(k. −λ) e−ik·x − d† (k. −λ)u(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. γ x (8. λ)v(k. op P (8.5) (8. (i˜ µ ∂µ − M ) ψ(˜) = 0. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). −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).

λ) e−ik·x + b† (k. i.19) Cop d(k. x (8.e. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. λ)C v T (k. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. (8. λ) Cop = ηC d(k. x (8. Top |φ = φt | = (|φt )∗ . λ). λ) eik·x . λ)u(k. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation.16) d k ηC d(k. The same relations hold for helicity states. . ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. λ)v(k.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). (8. λ) e−ik·x + b† (k. we start with the complex conjugated x Dirac equation for ψ. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.3 Time reversal xµ = (t. m).17) 3 = λ This shows that −1 Cop b(k. x x (8. As time reversal will transform ’bra’ into ’ket’. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . Norm conservation requires Top to be anti-unitary2 .22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ .18) (8. u(k. λ) −1 Cop = ∗ ηC b(k. m) ¯ = = v(k. The (time-reversed) Dirac equation becomes. antilinear operator is anti-unitary if A† = A−1 . λ). it is antilinear1 . x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0.Discrete symmetries 68 the latter being also a solution of the Dirac equation.15) (where one must be aware of the choice of spinors made in the expansion. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . m).21) where A is a 4 × 4 matrix acting in the spinor space. as discussed in section 4). r) −→ −˜µ ≡ −xµ = (−t.14) (8. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). m) ¯ C v T (k. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. x iγ µT ∂µ − M ψ(−˜) = 0. r). (2π)3 2Ek (8. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. λ)C uT (k. 8.

p.Discrete symmetries 69 Table 8. (8.28) (8.25) (check this for the standard representation). λ a |a.26) d k x x ηT b(k. λ| ¯ C |¯. λ) e−ik·x (2π)3 2Ek (8. λ) eik·˜ + d† (k. Since there are 16 independent 4 × 4 matrices. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . −λ |¯. −p. x (8. ∗ ˜ v (k. ˜ Top d(k. λ)u∗ (k. p. λ). λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ).32) (8.29) In table 2 the behavior of particle states under the various transformations has been summarized. λ| a.33) (8. λ). op T (8. p. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ) eik·x + d† (k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). ˜ = 0|b(k)|A 8. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k.34) .2: The transformation properties of physical states for particles (a) and antiparticles (¯). −p. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ)v ∗ (k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. p. a state |a.30) (8.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k. λ) eik·x + d† (k. −λ a T a.24) (8. λ)u∗ (k. λ)iγ5 Cv(k. λ). −p. The result is = = λ (8. λ a P |a. λ) Top = ηT b(k. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. λ) T −1 = η ∗ d(k. −p. λ) i γ5 C v(k. λ)v ∗ (k.31) (8. there are 16 independent of these bilinear combinations. λ)v ∗ (k. λ)u∗ (k. λ |¯. λ) eik·x + d† (k. λ)iγ5 Cu(k. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8.

(8.39) where Γµ (p′ . p′ |V µ (0)|p = u(p′ )γ µ u(p). p′ . p) can be built from any combination of Dirac matrices (1. as well as under the combined operation Θ = PCT (see Table 3). 2M (8.36) (8. γ ν ]. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . s′ |V µ (x)|p. γ5 γ µ or γ5 ).g. F. for the vector current V µ (x). p)u(p) e−i (p−p )·x . ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. As an example consider the vector current for a point fermion. 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 . Note that the lagrangian density L (x) → L (−x) under Θ. The 16 combinations of Dirac matrices appearing above are linearly independent. C.. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. in this case only q 2 = (p′ − p)2 .3: The behavior of the independent bilinear combinations of fermi ﬁelds under P.40) where the coeﬃcients are functions of invariants constructed out of the momenta. C. p′ |V µ (x)|p = u(p′ )Γµ (p′ . γ µ . σ µν . s . called form factors. ¯ In general the expectation value between momentum states can be more complicated. For instance for nucleons one has Γµ (p′ . 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 ). In many applications the expectation values of currents are needed. e. (8. ¯ ′ ′ (8. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . 8. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν )..37) (8.35) The x-dependence can be accounted for straightforwardly using translation invariance. of which the expectation value between momentum eigenstates can be simply found. F.5 Form factors Currents play an important role in ﬁeld theory. (q 2 ).41) 2M 2M .Discrete symmetries 70 Table 8.38) (8. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). (q 2 ). and T. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).

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

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

QH (t)|q. |ψS (t) = U (t. AS (t) = AS and the states o are time dependent. in which the operators are time-independent. t0 ). (9. (9.4) ∂ U (t. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t0 ) ∂t (9. in which the states are time-independent. [AH .3) (9.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t ′ |q . t ≡ q|q. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. ≡ 0. t . t0 ) AH (t0 ) U (t.1 Time evolution as path integral U (t. |ψH (t) = |ψH . (9. H].7) 73 . t0 )|ψS (t0 ) . t ′ QH (t′ )|q ′ .Chapter 9 Path integrals and quantum mechanics 9. AH (t) = U −1 (t. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .2) (ii) Heisenberg picture. t′ ≡ q ′ |q ′ . and the operators are time-dependent.5) (9. t0 ) = H U (t. Consider the two (time-independent) Heisenberg states: |q.6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing. (9. t′ .

q .. qj )]) . B]. (9. (9. The purpose of this is to calculate the evolution for an inﬁnitesimal time interval.13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential. .11) (9.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj . with the correction1 being of order (∆τ )2 . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . H(P. eA eB = eC with C = A+B+ (9. t′ |q. Then qj+1 . B] + [A.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . use the Campbell-Baker-Hausdorﬀ formula. . t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . 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|. The hamiltonian is an operator H = H(P.12) where the transformation between coordinate and momentum space involves q|p = ei p. which is a hamiltonian function in which the operators are replaced by real-numbered variables. B]] + [[A. |q1 q1 |e−i H∆τ |q . QS |q ′ ≡ q ′ |q ′ .8) Choose t as the starting point with |q = |q. qj ). ′ (9. ˙ 2π 1 1 1 [A. Q) = K(P ) + V (Q) = (P 2 /2M ) + V (Q) one can split e−i H∆τ ≈ e−i K∆τ e−i V ∆τ . . Combining the two terms gives qj+1 . 2 12 12 (9. 2π (9.15) By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj . . n |q (9. [A. tj+1 |qj ..14) .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|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 ′ .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . t′ |q. B] + . tj+1 |qj . t = q ′ |e−i H(t −t) |q . Q) expressed in terms of the operators P and Q. 2π (9.16) this.

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

Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . (9. Note that procedure (i) is an example of the more general procedure under (ii). N (9.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. the following integrals can be derived for a real symmetric or complex hermitean kernel K. A useful property of functional integration is the translation invariance. (9. α(x) = n αn fn and K(x. 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. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. 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 dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.25) (see Exercises). with Dα F [α] = n dαn N F (αn ) (9. Dα F [α] = Dα F [α + β]. 2 (9.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions.24) (9.21) again a multidimensional integral.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .27) (9.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.23) Having deﬁned the Gaussian integral. y) = ∗ m. . Consider the gaussian functional as an example. 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.

(9. Examples of functional diﬀerentiation are (9. i. θη = −ηθ. In the expansion of the exponential (from second to third line). (9.38) . θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. δα(x) δα(x) δ αKβ = dy K(x.34) (9. α(y) = δ(x − y). however. n θn fn (x) dx θ∗ (x) θ(x) . In principle the deﬁnitions of functionals is the same.32) (9.31) δ δ α = dy α(y) = dy δ(x − y) = 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. which vanish. the Gaussian-type functionals. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . θ2 = 0. F [θ.33) (9. δα(x) or in discretized form δ 1 ∂ ∂ with .30) (9.29) (9. e.35) For applications to fermion ﬁelds.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). (9. αy = δxy . −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . δα(x) (9. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. products of the same pair appear. αn = δmn .e. y) θ(y) .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. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 .g. dx dy θ∗ (x) K(x.

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

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

. T = φ0 (q.68) (9. T ′ |Q. Since we have (neglecting multiplicative factors). thus T →i∞ lim q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . Q(tn )|0 . .71) 9.70) for the generating functional is in fact ill-deﬁned.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′ |q. . dq dτ + J(τ )q(τ ) . 9. t′ ) q ′ . Better is the use ˜ of imaginary time t = −it. ˜ t′ → ∞. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . J=0 (9. 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.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. t). T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. . = (i)n 0|T Q(t1 ) .6 Euclidean formulation The above expression (Eq. δJ(tn ) hence the name generating functional. . t|Q.67) = = φ0 (q. J (9. (9. t 0 0out |0in J . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . t) (9.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. The importance of Z[J] is that the time ordered product of operators can be obtained from it.

.and Grassmann-valued functions. det K 1 . dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. δJ(tn ) = (i)n J=0 1 δ n ZE [J] .75) = − LE q. δ n Z[J] 1 Z[J] δJ(t1 ) . (9. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). Exercises Exercise 9. τ dq ˜ dt ≡ −L q. y) = ∗ m. (9.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions.g.2 Check the examples of functional derivation for real-. As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . + V (q). δJ(tn ) J = 0 ˜ t = it (9. complex. ZE [J] = = where LE q. when L q. 1 = √ .76) which is a positive deﬁnite quantity.n fm (x) Kmn fn (y). det K Exercise 9. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .73) d˜ LE q. for Grassmann valued functions. .74) − V (q) 2 (9. . The diﬀerentiations in Z[J] and ZE [J] are related. ˜ ˜ ZE [J] δJ(t1 ) . .77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. ˜ dt (9.72) (9. e. i dq . ensuring convergence for the functional integral ZE [J]. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. where Kp are the eigenvalues of the matrix Kmn in K(x. .Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.

80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .78) (9. y) K −1 (y.79) (9. . z) = δ(x − z). det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . 2 (9. where dy K(x.Path integrals and quantum mechanics 84 Exercise 9.

∂µ φ) + Jφ] (10. (10. .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. . . δJ(xn ) 85 ≡ G(n) (x1 .e. . x. . φ(x. . x′ . implying at the level of four vectors in coordinate and momentum space (when k · x ≡ kE · xE ) for instance x4 = i x0 = i t x · x = (x0 )2 − d4 xE = i d4 x L = 1 2 3 i 2 i=1 (x ) k 4 = −i k 0 =− 4 µ 2 µ=1 (x ) k · k = (k 0 )2 − d4 kE = −i d4 k 3 i 2 i=1 (k ) =− 4 µ 2 µ=1 (k ) ∂µ φ∂ µ φ − 1 2 M 2 φ2 LE = 1 2 µ (∂E φ)2 + 1 2 M 2 φ2 Furthermore we ﬁnd the n-point Green functions 0|T φ(x1 ) . t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . .3) The Euclidean formulation is as before. .4) J=0 . t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. t′ |φ. i.1) Dφ exp i σ d4 x L (x) and T′ Z[J] = lim T → i∞ T ′ → −i∞ Dφ exp − Dφ exp i T d4 x [L (φ. ∂µ φ) − Jφ] (10. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . xn ).Chapter 10 Feynman diagrams for scattering amplitudes 10. Π) (10.2) = ∞ −∞ d4 xE [LE (φ.

These Green functions appear in the expansion Z[J] = n in n! d4 x1 . . .10) J=0 In order to determine ∆F . . .8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. . As discussed in the previous section (exercise 9.6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. 2 2 → (10. (10.e.13) . xn ) J(x1 ) . . xn ) = x2 xn . . J(xn ). d4 xn G(n) (x1 . . (2π)4 (k 0 + iǫ)2 − E 2 (10. with Z0 [0] = 1. . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . (10. one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. Furthermore.3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . (10.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y).9) i. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. . . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 .

10.16) C where C is a closed contour in the complex k 0 -plane. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . i.e. µ 1 1 • Firstly. φ+ (0)] = (2π)3 2E are also shown in exercise (7. t) = 0 for t > 0.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x .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 2 − M 2 where the latter contour can be deformed to the contour for ∆C .20) = exp − 2 where or (see ﬁg. ∆F = ∆C . t) = 0 for t < 0. 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 . (10.14) They are solutions of the inhomogeneous equation. .22) (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. 4 k 2 − M 2 + iǫ (2π) (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. (2π)4 k 2 − M 2 (10. (2π)4 k 2 − M 2 (10.18) Thus we see various solutions. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).17) d4 k e−i k·x . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). (10. 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.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.4). 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 0 − iǫ)2 − E 2 (10. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. the advanced Green’s function ∆A (x) = satisfying ∆A (x.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). The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. ∆(x) = − d4 k e−i k·x .

φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1.24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. the same result as above. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . x2 .25) x1 x2 x3 = x4 + + . (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . . because this is the only quantity entering Z0 [J]. . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . x3 .3) i ∆F (x − y) = 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0 = i ∆C (x − y). We ﬁnd (see also section 7. we have G0 (x1 . φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). or diagrammatically (4) (10.Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. we 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. = θ(x0 − y 0 ) 0|[φ+ (x).

30) and in zeroth order G(0.28) 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 .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 .e. (10.32) (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. the generating functional can be written. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. x3 .34) Introducing a vertex point to which four propagators are connected.27) When interactions are present.g ) (x1 . x2 ) = i ∆F (x1 − x2 ) = G(4.Feynman diagrams for scattering amplitudes 89 10. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . x2 ) = G0 (x1 . (10. ≡ −i g d4 z (10. x3 . x2 . x2 . with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[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. x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.g 0 ) = G0 = 1.g ) (x1 .35) z . i.31) G(2. (2) (0) (10. Z[J] = = Dφ exp i exp i d4 z LI (10.29) This expression will be the one from which Feynman rules will be derived. x4 ) = G0 (x1 . (10. L (φ) = L0 (φ) + LI (φ).

g 1 ) = 1 8 z 1 8 . . did not contain parts that could be written as products of simpler Green’s functions. J(xn ) c (10.37) G(2. .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. . .42) x1 G(n) (x1 . d4 xn G(n) (x1 . .39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . . .g ) (x1 . . .38) The result for the 4-points Green’s function is left as an exercise. . i. (10. Z[J] = n in n! d4 x1 . In the free case we have seen that the exponent of Z[J] contained all essential information.36) J J from which one obtains G(0.e.40) in−1 n! d4 x1 . Connected Green’s functions We have now seen the Green’s functions as the quantities appearing in the expansion of the generating functional Z[J]. . . Diagrammatically this exponent only contained the two-point function. which also was the only Green’s function for which the diagram was connected. . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . xn ) J(x1 ) . J=0 (10. . . J(xn ). . . . G(n) (x1 . . . xn ) J(x1 ) . (10. . . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . δJ(xn ) . . . This remains also true for the interacting case. d4 xn G(n) (x1 . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . (10. To see this write Z[J] = exp (i W [J]) with (by deﬁnition) the expansion W [J] = n or i W [J] = ln Z[J]. (10. .41) in terms of socalled connected Green’s functions. (10.

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

out|p′ . in|S = α. x1 G(4. this will be translated to the action on ﬁelds. in = |α. implying the absence of tadpoles. the vacuum expectation value of the ﬁeld φ(x) is zero. . in = p. in|S = 0.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. out = p|p′ . p′ . out . Proof: α. Proof: 0.g ) (x1 . then (n) (n) Gsc = Gc for n ≤ 3. in|S|p′ . . (2) (10.e.45) x2 as the only surviving diagram (see exercises). out . thus β. x3 . out| = eiϕ0 0. ∂µ ∂ µ + M 2 φin (x) = 0. in = |p1 . Sβα ≡ β. out| and S † |α. e. in into ﬁnal state particle states (out-states) |β.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. x4 10. out|p′ . (10. in = p. φout and the interpolating ﬁeld φ(x). pn .47) . . in|p′ . p. in = β.Feynman diagrams for scattering amplitudes 92 i. . 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 . translation invariance). For the free scalar theory. (10. . x4 ) = c 1 x3 z (10. considered explicitly. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . in ⇔ β. (2) The one-particle state is invariant (conservation of energy and momentum. in|SS † |α. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). . x2 . Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . out|α. one ﬁnds for the connected 4-point Green’s function at order g. in = S|α. (10. out|α.44) J=0 For the interacting theory. out (suppressing except momenta all other 1 m quantum numbers). Next. out = δαβ ↔ SS † = 1. out = |p′ .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| (choose ϕ0 = 0).g. in = p. in|S = β. out| ⇔ |α.

S is invariant under e Poincar´ transformations: U (Λ. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S.e. 9. (10. β.and out-ﬁelds is: φin (x) = S φout (x)S −1 . 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10. (2π)3 2E (10. The convergence therefore must be weakened to t→−∞ √ (10. in|φin S → φin S = S φout (x). a causality condition that can be proven to imply the absence of interactions. √ Although the above. The relation between S-matrix and in. the source term Jφ in the lagrangian density is treated in the interaction picture.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. The relation between S and Z[J] To establish this relation. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . z) φ(z) d4 y d4 z ∆A (x − y) K(y.54) . this can actually not be used as it would imply [φ(x).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).g. a) = S. (10. The time evolution operator (see Eq. φin (y)] = iZ ∆(x − y). e.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.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. Proof: β. 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. implies the strong (operator) convergence φ(x) → Zφin (x). z) φ(z). which in a later stage (renormalization) will become more important. a)SU −1 (Λ. i.51) (10. as it stands. φ(y)] = Z [φin (x). in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. 2 ∂µ ∂ µ + M0 φ(x) = J(x).52) where Z is a constant.Feynman diagrams for scattering amplitudes 93 while φ satisﬁes the inhomogeneous Klein-Gordon equation with the bare mass M0 (this is the mass appearing in the lagrangian L0 ). (10.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . out|φout = β.

e e ] = [A. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. (10.62) where F [J] is some (arbitrary) functional. B] (for the case that [A. we can. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x.63) . y) δ δJ(y) : F [J]. 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. φ(x).56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. C] are c-numbers. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. z) . express it in terms of advanced and retarded Green’s functions. [A. 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. φ(y)] f (y) e e . t i Ut∞ [J] φ(x)U−∞ [J] (10.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. eB ] = [A. 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). just as we did for φ(x).57) i [φin (x). : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.61) e φ+ f where the normal ordered expression : e φ f : is used. B]eB . S U [J]] = √ Z δ S U [J]. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. B] and [A.55) δU [J] δJ(x) = t→∞ t→−∞ −→ −→ √ i φ(x) U [J] → i Z φout (x) U [J] √ i U [J] φ(x) → i Z U [J] φin (x). Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). δJ(z) (10. e = i. z) δU [J] . δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10.59) (10. B + C]e e . (10. one has [A.58) In order to ﬁnd a solution to this equation note that. z) d4 y d4 z ∆(x − y)K(y. φ− f (10. δJ(z) (10. provided [A. which is equal to the expression e in which the creation operators are placed left of annihilation operators. B] is a c-number).e.

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

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

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

e. η]. 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 . but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams. namely (Rule 5) Feynman diagrams which only diﬀer by exchanging identical fermions in initial or ﬁnal state have a relative minus sign. (Rule 6) Each closed fermion loop gets a sign −1. in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is (see next section for e-e-γ vertex)..Feynman diagrams for scattering amplitudes 98 Dirac ﬁelds For fermions the generating functional is given by Z[η. The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution.− i δη(z) i δη(z) d4 x d4 y η(x) iSF (x − y) η(y) .g. 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. − g2 2 dz dz ′ δ2 δ2 δη(z) δη(z) δη(z ′ ) δη(z ′ ) and the quadratic term in Z0 [η.e. (10. η).. which arises from the quadratic term in exp i d4 z LI . which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. = + .71) = exp i where iSF is the Feynman propagator for fermions. . 4 2 − M 2 + iǫ (2π) p and the (oriented) propagator in Feynman diagrams involving fermions is i p j (10.72) (10.. − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ ). η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ .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.

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

s3 )γ µ u(k.76) 1 .s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .s3 . To lowest order (∝ α = e2 /4π) only one diagram contributes. s1 ) u e2 u(p′ .5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )].79) . s3 )(−ie)γ µ u(k. s′ ) ¯ u s. t2 (10.Feynman diagrams for scattering amplitudes 100 10. s)γν u(k ′ .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) (10.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.s4 e4 (m) µν (M) L L . s2 )|2 ¯ q2 (10. s′ )γµ u(k. s4 )γµ u(p. ¯ q2 (10. where u ¯ L(m) µν = = 1 2 u(k ′ .s2 . s)¯(k. s2 ).s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(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 ′ . s1 ) ¯ −igµν u(p′ . 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 . s4 )(−ie)γ ν u(p.

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

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

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

12) pb = (mb . m2 ). 2mb (11. for instance. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . (11. m2 ) = 4(pa · pb )2 − 4p2 p2 .8) (11. 4s (11. p1 . In that case trf pa = (Ea . m2 .11) (11. ptrf ). trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b .10) The function λ(s.13) (11.14) One can. 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 . p2 ) = = where λ12 denotes λ(s. m2 ) a b . Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . which in the speciﬁc case a b also can be expressed as λ(s.16) . 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 √ = .15) Explicit calculation of the reduced two-body phase space element gives dR(s. use the ﬁrst relation and the abovementioned threshold value for s to calculate the threshold for a speciﬁc n-particle ﬁnal state in the target rest frame. m2 ) a b . m2 . = 2 s λ(s.9) (11.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . p1 . 2mb λ(s. m2 . 4π s 4π 8π s 4π (11. 0). Also in this case one can express the energy and momentum in the invariants. mb and s). a (11. m2 . 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. m2 ) is a function symmetric in its three arguments. 2 s s − m2 + m2 a b √ .

H1 + H2 → h + X. Of course there are not only 2-particle channels. usually many particles are produced. pc . in exclusive measurements all particles are detected.. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels.4).18) (11. (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees.. Examples appear in → π− + p (11. = (11. py ).24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. E . The variable s is always larger than a c b d the minimal value (ma + mb )2 .21) (11. At high energies.27) with m2 = m2 − p2 = m2 + p2 + p2 .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. tmax min λcd /4s.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . Instead of the scattering angle. usually is a 2-particle state. 2s (11.17) (11. Consider the process a + b → c + d. 11. there is usually a preferred direction.22) (11. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. pd ) = 4π s 4π 8π s 4π dt dt √ √ . The initial state. however. in which one particle is detected. pz = mT sinh y.20) → π0 + n → π+ + π− + n → . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . . s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. Consider the 1-particle inclusive case. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). An often used set of invariants are the Mandelstam variables. 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 . Writing E = mT cosh y. In inclusive measurements no particles are detected in the ﬁnal state.25) dR(s. 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 .19) (11. The various possibilities are referred to as diﬀerent reaction channels. pT = (px . A speciﬁc reaction channel starts contributing at the threshold value (Eq.

¯ b (11. (11. tt = s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). t. u = (p1 − p4 )2 = (p2 − p3 )2 . ut = u) = Mab→cd (s. t. η = − ln(tan(θ/2)) = tanh−1 (cos θ). With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1.31) . 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. 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). uu = s) = Mab→cd (s. 11. b ut = (pa − pd )2 = u.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β.28) which (only involving angles) is easier to determine. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. tu = s. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. u).2 Crossing symmetry In the previous chapter. c bd Mad→c¯(su = u. ¯ tt = (pa − p¯)2 = s. (11. This is known as crossing symmetry. t = (p1 − p3 )2 = (p2 − p4 )2 . uu = (pa − p¯)2 = s. t.30) (11. ¯ tu = (pa − pc )2 = t. which means that rapidity distributions dN/dy maintain their shape. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. t-channel and u-channel processes respectively. u) with s = (p1 + p2 )2 = (p3 + p4 )2 .

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

Scattering theory 108 Although this at ﬁrst sight does not look covariant.36) (11. (11.34) i. it is. pn ).39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). σ= √ 2 λab T · V (11. N 1 . N and T · V are covariant. as expected. dt (11.35) This quantity is not covariant. (2π)4 δ 4 (Pi − Pf ) 2 = (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf ) d4 x ei(Pi −Pf )·x V. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. . 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. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 .37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears. . . the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . For the (proper) decay width one thus has 1 N Γ0 = .e. 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 .33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). T · V ρa (11. . . (11. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ.T V.38) (in which we can calculate −iMf i using Feynman diagrams). 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 .

44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. 2 λab and for instance for two particles dσ = q ′ M (s. . respectively. p1 . q n ) q n qn √ Mni (q i . θ) or M (q i . . pn ). The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. θcm ) √ q 8π s 2 2 (11. 11. implies for the scattering matrix M . . (11.42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . . p1 . . θ) in quantum mechanics.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM.e. pn ). q f ). (11. The result is N = |Mf i |2 dR(s. θ) = −8π s ℓ (in analogy with the expansion for f (E. θ)|2 . 8π s 2π 8π s .46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. i. ˆ ˆ Inserted in the unitarity condition for M . q n ). .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f .40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . (11.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . p1 . note the sign and cos θ = qi · qf ). t) λab 4π dt. θ). . (11. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). .Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). . for which one has dσ/dΩ = |f (E.43) dΩcm = π M (s. for which the √ amplitudes squared have been calculated in the previous chapter. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. . δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i . . 16π 2 s (11.47) M (s.41) (11. (11. see 11.

2i q 2i q (11.51) and for the case that this is the only channel (purely elastic scattering. Using σ= dΩ q ′ M (s.Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . In general a given channel has |Sℓ (s)| ≤ 1.52) From the imaginary part of M (s.54) ℓ . 0). q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni .49) or Im Mℓ = q |Mℓ |2 . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . n (11. ℓ (11.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . (11. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). the total cross section can be determined.e. (11.53) The diﬀerence is the inelastic cross section. θcm ) √ q 8π s 2 . Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). q ˆ i.50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). (11. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . (11.

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

e. M2 − s (11. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. At resonance the cross section σT (π + p) is about 210 mb. Γ = 2. This characteristic shape of a resonance is called the Breit-Wigner shape. This is a narrow resonance discovered in 1974. the full width is Γ = 88 keV. where furthermore σinel = 0. u) + g(t. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . All these decays can be seen and leave an ’invisible’ width of 498 MeV.4) can be written as 4πα2 dσ = {f (t. The half-width of the resonance is M Γ. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . t)} . which is attributed to neutrinos. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. Knowing that each neutrino contributes about 160 MeV (see next chapter). The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. Exercises Exercise 11.49 GeV. the partial width into e+ e− is Γee = 5.88 MeV.1 Show that the cross section for electron-electron scattering (exercise 10. • The J/ψ resonance in e+ e− scattering. 2 q (s − M 2 )2 + M 2 Γ2 (11. u) = g(t. Its mass is M = 1232 MeV.26 keV. a fast change in the phase shift for a narrow resonance. its width Γ = 120 MeV. Limiting ourselves to a resonance in one channel. Its mass is M = 3096.Scattering theory 112 (Prove this using the unitarity condition for partial waves). • The Z 0 resonance in e+ e− scattering with M = 91. i. u) + f (u. dt s(s − 4m2 ) with f (t. Three neutrinos explain the resonance shape. The cross section at resonance is about 30 nb.2 GeV.59) reaching the unitarity limit for s = M 2 . t) + g(u. Note that because of the presence of a background the phase shift at resonance may actually be shifted.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. 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).

Scattering theory 113 Exercise 11. Calculate now fπ .999 877 Γ and Γ(π − → e− νe ) = 0. a quantity which we will ¯ ¯ encounter in the next chapter (section 12.8. cm |va − vb | in the center of mass system for two colliding particles a and b. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.2). Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. Exercise 11. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) .4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ .000 123 Γ. 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. Compare the results with the experimental π − lifetime τ = 2. 4. Exercise 11. ρcm ρcm a b 1 cm .2 Calculate the ﬂux factor. As discussed for unstable particles. writing N (s) M1 (s) = 2 + iM Γ .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0.

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

12. In principle such a phase in a given point is not observable. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed.15) ψ(x). should be independent of such transformations. However. (12. -dx -dy µ µ dyµ µ dx For a vector.19) This gives rise to a (path dependent) phase in each point. the real change thus is given by ψ(x) = (1 − dy ρ Dρ )(1 − dxσ Dσ )(1 + dy ν Dν )(1 + dxµ Dµ )ψ(x) = 1 + dxµ dy ν [Dµ . ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).e. 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.16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx).16. which connects two identical vectors. (12. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).The standard model the basis for x G and x+dx 117 G.e. The relation in Eq. but expresses them with respect to diﬀerent bases. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. (12. (12. ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0).14) (12.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). i. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. = . ds which is solved by dψ(s) ds = dxµ . i. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.

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

The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. for the vacuum Qa |0 = 0. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa .e.23) F +η the latter only forming a minimum for m2 < 0. . λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). while the quadratic terms must correspond with non-negative masses. . A quantum ﬁeld theory has only one nondegenerate groundstate |0 . It is only invariant under rotations around the 3-axis. etc. Any constant ﬁeld with ’length’ F is a possible groundstate. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. .) terms constitute interaction terms. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. In the case of degeneracy. They are known as the Weyl mode or the Goldstone mode: Weyl mode. In this case the spectrum is described .25) 2 2 2 Therefore there are 2 massless scalar particles.The standard model 119 a constant ﬁeld (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2. In this situation one speaks of spontaneous symmetry breaking. The situation now is the following. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. (12. therefore a choice must be made. The η massless particles are called Goldstone bosons.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value.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. ϕc = 0|φ |0 = (12. say 0 0 . (12. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . corresponding to the number of broken generators (in this case rotations around 1 and 2 axis) and 1 massive scalar particle with mass m2 = −2m2 . (12. The classical groundstate appears degenerate. i. has less symmetry. 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. ϕ2 and η. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . while the lagrangian including the nonzero groundstate expectation value chosen by nature.

ψu mu md ψ = ψd .The standard model 120 by degenerate representations of the symmetry group. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. if the generators Qa generate a symmetry. Goldstone mode. . Neglecting at this point the local color symmetry. there must be a nonzero expectation value 0|Jµ (x)|π (k) . denoted |π a (k) .29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor).) one can write L = f ψ f (i/ − Mf )ψf . V µ = ψγ µ T ψ. (12. ψ −→ ei α·T ψ. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . Using the Dirac equation. (12. a and a′ are degenerate states. π (12. (12.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space.e. i. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . ∂ ∂ (12. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. This means that they are operators that create states from the vacuum. .26) for all the states labeled by a corresponding to ’broken’ generators. T ] ψ. i.. . called ﬂavors. etc. .e. irrespective of the fact if they annihilate the vacuum. and one must have mπa = 0. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m).27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab (fπ = 0) (12. T = τ /2. The generators Qa (in that case the rotation operators Lz .31) . it is easy to see that one gets ∂µ V µ = i ψ [M. H] = 0. A bit more formal. Including a sum over the diﬀerent ﬂavors (up.32) (12. 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.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators.e. [Qa . a massless Goldstone boson for each ’broken’ generator. = ψ(i/ − M )ψ. strange. down. i. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). M = . Taking the derivative. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . a ∂ µ Jµ = 0.

SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. parity would not play a role in the world. i. is by nature not realized in the Weyl mode. as the quantity ∂µ Aµ (x). (12. if in ﬂavor space M is proportional to the unit matrix. it is easy to see that one gets ∂µ Aµ = i ψ {M.3 MeV/c2 and Mn = 939.e. T } γ5 ψ. 12.e. This situation. which is approximately true for the up and down quarks. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. but also a triplet (isospin 1) of pions.27. The chiral ﬁelds ψR and ψL are transformed into each other under parity. (12. ∂ ∂ (12. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. parity minus). which is equivalent 2 to the V-A symmetry. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . e. chiral symmetry is not perfect. much smaller than any of the other mesons or baryons. T γ5 = τ γ5 /2.6 MeV/c2 ). proton and neutron. M = m · 1. i. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. (12. gives also rise to a nonzero mass for the Goldstone bosons according to Eq.e. ψ −→ ei α·T γ5 ψ.e. In the real world. with almost degenerate masses (Mp = 938. (leaving out the ﬂavor structure) the same as ψγ5 ψ. while the groundstate is only invariant under isospin rotations (R = L). This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry.g. but the fact that the symmetry is not perfect and the right hand side of Eq.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . Therefore the world of up and down quarks describing pions. a doublet (isospin 1/2) of nucleons. How can we see this. socalled axial vector transformations. Note that these transformations also work on the spinor indices. From group theory (Schur’s theorem) one knows that the latter can only be true. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . behaves as a pseudoscalar particle (spin zero. however. 12. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR .35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. the lagrangian density is invariant under left (L) and right (R) rotations independently. stated equivalently. i.35 is nonzero. . generated by TR/L = 1 (1 ± γ5 )T . We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. etc. for which the ﬁeld (according to the discussion above) has the same behavior under parity. This symmetry. I. both are very small. 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. etc.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. is what happens in the real world. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. including in Dirac space a γ5 matrix. There exists another set of symmetry transformations. Aµ = ψγ µ T γ5 ψ. This combined symmetry is what one calls chiral symmetry. ψR/L = 1 (1 ± γ5 )ψ. From the number of broken generators it is clear that one expects three massless Goldstone bosons. fπ = 0 remains.e. or.34) Using the Dirac equation. where the quark masses are not completely zero. T ] = 0.

thus 9 independent degrees of freedom. 0 ϕc = 0|φ |0 = 0 .40) a Dµ φ = ∂µ φ − ig Wµ La φ. (12. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation.42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). 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 + . The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle.The standard model 122 12. (12. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. like a photon. We have the possibility to perform local gauge transformations. After symmetry breaking the same number (as expected) comes out. in this case three-dimensional) representation reads (La )ij = −i ǫaij . . having two independent spin components) and three scalar ﬁelds (one degree of freedom each). One may wonder about the degrees of freedom.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. 0 φ= (12. made into a gauge theory. 4 2 where Since the explicit (adjoint.37) (12. again 9 degrees of freedom.39) (12. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . Dµ φ = Gµν = ∂µ φ + g Wµ × φ. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. The diﬀerence comes when we reparametrize the ﬁeld φ. .e. (12. The symmetry is broken in the same way as before and the same choice for the vacuum.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 . Initially there are 3 massless gauge ﬁelds (each. is made. i. 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). . but one has 1 massless gauge ﬁeld (2). 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1).38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . . while the number of massive gauge bosons coincides with the number of ’broken’ generators. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). 0 0 0 = . as in this case there are no massless Goldstone bosons.

muon µ− or tau τ − ).45) (12. 0 1 +1/2 L νe 3 − = with TW = and TW = L= −1/2 eL L 2 R = R − = eR 3 with TW = 0 and TW = 0. explicitly L −→ e−i β/2 L. U(1)Y U(1)Y (12. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. (12. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . the quarks (up. R −→ e−i β R. electromagnetic and weak forces.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.e. its corresponding neutrino (νe .44) which has an SU (2)W symmetry under transformations eiα·TW . 2 (12. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.48) (12. τ ). ψR/L = 2 (1 ± γ5 )ψ decouple. neutrinos. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 .and righthanded particles. down.47) and YW is considered as an operator that generates a U (1)Y symmetry. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. (12. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. 1 As they are massless. etc. while the lefthanded particles form a doublet.50) (12. left. that gives a good description of the physical world. i.The standard model 123 12. This is written as 3 Q = TW + YW . muons. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment.) and the gauge bosons responsible for the strong. however.). Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. νµ and ντ ) and their antiparticles. The procedure. etc.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet.51) . the leptons (elektrons.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.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. 2 2 = ∂µ R + i g ′ Bµ R. R. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. ∂ ∂ ∂ (12. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. = ∂µ L − (12.

† 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. 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. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. . (12.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . + 8 (12. (12. h) (12. leading to the parametrization 1 0 φ(x) = √ (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 .53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. . 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) .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .58) (12. Putting this in leads to L (f ) =L (f 1) +L (f 2) . For the classical ﬁeld the choice θ4 = v is made. 4 4 In order to break the symmetry to the symmetry of the physical world. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . . −V (φ) L and (h2) = −Ge (LφR + Rφ L). . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). the U (1)Q symmetry (generated by the charge operator). this leads to the lagrangian L (h1) .59) .56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms. (12. Using local gauge invariance θi for i = 1.

62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .60) (12. 2 Zµ (12.69) . 2 (12. (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12. we can read oﬀ e = g sin θW = g ′ cos θW .The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.63) (12. 4 2 MW g2 v2 = . 2θ 2θ 4 cos W cos W 0. 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 + .64) (12.65) The weak mixing angle is related to the ratio of coupling constants.61) (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 . 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 .66) From the coupling to the photon. (12.67) The coupling of electrons or muons to their respective neutrinos.68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (µ)† (jL )ρ (e) (jL )ρ GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ . . g ′ /g = tan θW .

(12. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. Finally we want to say something about the weak properties of the quarks. 3 = TW . Comparing this with the total width into (invisible!) channels. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV. (12. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). The coupling to the Higgs particle is weak as the value for v calculated e.76) with CV CA 3 = TW − 2 sin2 θW Q.231 2. one has the relation g2 e2 1 GF √ = . 2 2 2 (12. . Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three. i. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.74) (12.5 MeV (exp. muon or tau.80) First. from the MW mass is about 250 GeV.77) (12. i.g.e. giving rise to a running coupling constant after renormalization of the ﬁeld theory.40 GeV. 80. g2 MZ 2 2 (CV + CA ). such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.71) GeV −2 = = = = . (12.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions.166 4 × 10 0.19 GeV. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. v (12.75) The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with 3 TW 1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 .73) (12. (12. −5 (12. With the choosen vacuum expectation value for the Higgs ﬁeld. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). g ′ and v determine a number of experimentally measurable quantities. 91.e. me /v is extremely small. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. 1.72) (12.

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

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. each of these containing the three families.22 and η ≈ 0. i.g.86) the unitary CKM-matrix introduced above in an ad hoc way. . 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 . respectively). The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . 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 .82) space. u u where we include now the three lefthanded quark doublets in QL . The Higgs ﬁeld is still limited to one complex doublet. d d (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. A ≈ 0. † Λu Λ† = Vu G2 Vu . the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . (12. (12. the weak mass eigenstates are L (h2.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. 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 12.83) where Vu and Vd are unitary matrices.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. ms and mb ).82 and ρ ≈ 0.34.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . e.e. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. UR = uR cR tR . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .227. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .

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 . As before. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER .91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. UPMNS a parametrization that in principle also could have been used for quarks. In this case. one can write Λe = Ve Ge We and we ﬁnd † L (h2. = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12.e. we ﬁnd massless neutrinos. decoupling from all known interactions. and obtain L (h2. since the weak current is the only place where they show up.89) with three righthanded neutrinos added to the previous case. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . (12. mµ and mτ . (12. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . 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. EL = Ve EL . e N L= L EL (12.92) . † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . i.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL .ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . (12. The mass ﬁelds ER † mass † mass = We ER .ℓ) = −LφΛe ER − ER Λ† φ† L.5×10−3 eV2 . there is no family mixing for massless neutrinos. 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.87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. Working with the mass eigenstates and the weak current becomes EL γ NL = EL e n L weak for the charged leptons we see that the weak eigenstates for the neutrinos are NL = Ve† NL with the relation to the mass eigenstates for the lefthanded neutrinos given by † ′ weak mass mass NL = NL = Ve† Vn NL = UPMNS NL . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. The mixing is intriguingly close to the Harrison-Perkins-Scott tri-bimaximal 23 mixing matrix 0 0 2/3 1/3 0 2/3 1/3 0 1 1/2 − 1/2 − 1/3 UHPS = 0 1/3 − 1/2 . (12.

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

related via c Υ1R = U nL . i.e. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. 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′ . (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 .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 . Υ2R nR (12.105) for i = 1. T ] ψ. Prove that if the generators Qa generate a symmetry. a and a′ are degenerate states. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . Exercise 12. L Υ2L nR Υ1 and Υ2 .The standard model 131 taking ML and MR real and non-negative. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . H] = 0. (12. Exercise 12. . 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). T } γ5 ψ.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)) .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 . Exercises Exercise 12. i ψ {M. (12.1 Consider the case of the Weyl mode for symmetries.2 Derive for the vector and axial vector currents. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M.e. 2 U (M M † ) U † = M0 . i. [Qa . 2 with real masses Mi .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .

In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) .231. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ).5 Show that the couplings to the Higgs (for W + W − h. (b) Calculate the width to electron-positron pair. 3 TW . and the width to a pair of neutrino’s. mb ≈ 5 GeV and MW ≈ 80 GeV. ZZh. Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. the weak mixing angle is given by ¯ sin2 θW = 0. at least up to a high (which?) order in λ. The mass of the Z 0 is MZ = 91 GeV. Exercise 12. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . Γ(Z 0 → e+ e− ). hhh. 2 2 The masses are mt ≈ 175 GeV. .4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW .7 In this exercise two limits are investigated for the two-Majorana case. Mf M2 and ghf f = . etc. Exercise 12.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. e+ e− h. Γ(Z 0 → νe νe ). Exercise 12. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 .) are proportional to the mass squared for bosons or mass for fermions of the particles. Exercise 12. 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 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.05 eV. Given that neutrino masses are of the order of 0. Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 . 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 . (b) A more interesting situation is 0 = ML < |MD | ≪ MR .