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

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

November 2011 (version 6.04)

Contents

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

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

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

1

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

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

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

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

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

Peskin and D. Schroeder. 2007. Cambridge University Press.3 3.2 2.5 5. I: Foundations.3 6.8 3. Cambridge University Press.3 3.3 4. Peskin & Schroeder and Srednicki. Vol. 3.2 3. 2. II: Modern Applications. 1980.4 4. 3.2 3. 1. M. Zuber.3 2. 1995.2 1.1 3. but are rather compact.4 3.2 36 22 . 1994. These notes Section 1. An introduction to Quantum Field Theory. 2.5 2. McGraw-Hill.1 2.3 2.E.4 2. Quantum Field Theory. Diagrammatica. Ryder.2 4. L.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. Cambridge University Press.4 6. Srednicki.1 5.3.1 2.H. Cambridge University Press.4 3.1 2 2 3.2 3. Other text books of Quantum Field Theory that are useful are given in refs [3-6].1 6.2 3. The quantum theory of ﬁelds. Quantum Field Theory. S.5 2.V.3 3. Cambridge University Press. Addison-Wesly.5 4.1 4. Weinberg.1 3.2 6. 1996.3 2.2. 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). M.3 3. 4.4 2.-B. 1985. Vol.2 2. Veltman.7 2. M.4 3.3 References As most directly related books to these notes.3 2.2 3.3 3. Quantum Field Theory.2 2.1 1. 6. 1995.2 6. 3. Itzykson and J.1. Corresponding chapters in books of Ryder. The notes contain all essential information. 5.3 2. C. I refer to the book of Srednicki [1] and Ryder [2].

4 7.2.1 20.10 8. 6.6 9.1 20.3 12.2.5 6.3.5 .8 6.3 11.3 3.5 7.3.2.1 10. 4.1 8. 9.4.5 2.5 4.4 10.7 4.2 12.3 6 8 9 Srednicki 3 3 3 37 5.7.1 20.4 12.3 2.2 11. 6.3 8.3 6.2 5.2 9.6 8.4.5 6 7 6 6.5 11.4.6 3.3 8.2.3 7.3 7. 2.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.5 9.6 6.1 12.2 8.4 12.1 7.1 11.2 8.4 9.1 9.4 4.5 4.7 5.1.6 3. 2. 6.5. 4.1.1.3.5 9.4 3. 9. 4. 20.6 10.8 5 4. 6.1 9.3 10. 8.5 7.4 8.3 20.2 4.6 3.2 10. 4.4 2.3 9.6 4.2 7.4 These notes Section 7. 6.1 5.

5 Corrections 2011/2012) • Extension of Exercise 3.3 • First two sentences of Chapter 5 have been modiﬁed.2 . • Slight rewriting of discussion of Majorana mass term in section 6.

(1. Corresponding position and momentum operators do not commute. Indeed. S ≫ . ˜ ψ(p) = i d3 r exp − p · r ψ(r). In the framework of classical mechanics as well as quantum mechanics the position of a particle is a well-deﬁned concept and the position coordinates can be used as dynamical variables in the description of the particles and their interactions. In quantum mechanics. (1.582 119 15 (56) × 10−22 MeV s. (1.2) In the limit that v ≪ c one reaches the non-relativistic domain. being eigenvalues of the position operators. (1. (1.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. quantum eﬀects do not play a role anymore and one is in the classical domain. pj ] = i δij . c = 299 792 458 m s−1 . 1 (1.Chapter 1 Introduction 1. the action of the momentum operator in the coordinate representation is not as simple as the position operator.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. One can talk about states |r and the wave function ψ(r) = r||ψ .4) where δij is the Kronecker δ function. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). usually given divided by 2π. the position can in principle be determined at any time with any accuracy. It is given by pop ψ(r) = −i ∇ψ(r).054 571 68 (18) × 10−34 J s = 6. ≡ h/2π = 1.6) .1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united.1) In the limit that the action S is much larger than . They satisfy the well-known (canonical) operator commutation relations [ri . In quantum mechanics a special role is played by Planck’s constant h. In special relativity a special role is played by the velocity of light c.

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.

Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl .. • ds2 = 0 (lightlike or null intervals).. however. The lower indices are constructed in the following way. The real.3) with basis nµ (µ = ˆ 0. xi are the three space components. x1 . (1. The length (squared) of a vector is obtained from the scalar ˆ product. δim δjm δkm δin δjn δkn . We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices.νn can be conν structed. x3 ). One can distinguish: • ds2 > 0 (timelike intervals). independent of a coordinate system.3).µn = Λµ1ν1 . . are . Many other four vectors and tensors transforming like T ′ µ1 . The Poincar´ transformations include e e Lorentz transformations and translations. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). x2 . • ds2 < 0 (spacelike intervals). such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 .25) Within Minkowski space the real. So we could have used all upper indices in the above equations.21) (1. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. x2 .. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . x3 ) = (t.. xµ = gµν xν . One has x2 = xµ xµ = t2 − x2 . When 3-dimensional space is considered as part of Minkowski space.24) The quantity gµν ≡ nµ · nν is the metric tensor. In Minkowski space.2. linear transformations that do not change the length of a four-vector are called the Lorentz 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 . in which case zi = ǫijk xj yk . 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 . Vectors are denoted x = xµ nµ . Λµn n T ν1 . 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. one must distinguish tensors with upper or lower indices and one can have mixed tensors. The (invariant) lengths often have special names. The scalar product of two diﬀerent vectors x and y is denoted x · y = xµ y ν gµν = xµ yµ = xµ y µ = x0 y 0 − x · y. and are given by (x0 .22) (1. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . denoted as V ′µ = Λµν V ν .23) δjm δkn − δjn δkm .26) (1. the points lie on the lightcone and they can be connected by a light signal. ˆ ˆ (1. Unlike in Euclidean space. These transformations do change the components of a vector. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. it is convenient to distinguish lower indices from upper indices. the invariant length or distance (squared) is not positive deﬁnite. −x). In special relativity we start with a four-dimensional real vector space E(1.1. . where the coordinate t = x0 is referred to as the time component. Relations relating tensor expressions. x) = (x0 . we will use upper indices for the three-vectors. No diﬀerence is made between upper or lower. x1 .Introduction 5 that can be used in the cross product of two vectors z = x × y. (1. Because of the diﬀerent signs occurring in gµν .

Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.32) . ∂1 . The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . deﬁned by ∂2 − ∇2 . but more important has the same eﬀect on lefthandside and righthandside. ∂2 . 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. not aﬀected by Lorentz transformations. . where m is called the mass of the system. Each of these permutations leads to a minus sign.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.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used).35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . explicitly written as (p0 . = ∂ . (1. (1. a∞ = 4πǫ0 2 /me e2 . ∂t (1. Note that in this equation one has on left. g0 = g1 = g2 = g3 = 1.∇ . (1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . for instance.28) ν ν Note that gµ with one upper and lower index. because one has (1.27) It is easy to convince oneself of the nature of the indices in the above equation.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. 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. ∂3 ) = ∂ ∂ ∂ ∂ . ∂t ∂x ∂y ∂z ν ∂µ xν = gµ .36) Exercises Exercise 1. p) = (E. . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. p).33) (1. . (1. constructed via the metric tensor itself. (1. = −24. It is an invariant tensor.Introduction 6 called covariant. The length squared of ∂ is the d’Alembertian operator.34) (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 . ′ ′ . starting from ǫ0123 = 1. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 .

Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . ˆ ˆ ˆ ˆ These are n1 = (0. Compare it with the proton mass and use Eq. a+ . a+ . a1 . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . . n1 . e. 0) ˆ n3 = (0. 0) ˆ n2 = (0. me . (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). n3 . 1) ˆ . 0) ˆ n0 = (1.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). n1 . a1 . 1. a1 . 0. 0. [Note: what is the MKS unit for V/T?] Exercise 1. g+− and g−+ . a3 ) are the expansion coeﬃcients using the basis vectors n0 .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . ǫ0 . 1. ν. a2 . n2 . a2 ] we can ﬁnd basis vectors n− .2 Prove the identity A × (B × C) = (A · C) B .13 to give its actual value in kg. One does not need to know . but only appropriate combinations. 2. . ]. 0. 1. a2 ] or [a− . (b) The coordinates (a0 . n+ . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c.(A · B) C using the properties of the tensor ǫijk given in section 1. ˆ ˆ . 0. ρ. 0. How are a+ and a− related to a+ and a− . α and me c2 = 0.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . n2 . by a simple few-line reasoning [For instance: If {µ. Exercise 1. which is the Compton wavelength for the Planck mass. Exercise 1. Also for the coordinates [a− . . g−− . This demonstrates how a careful use of units can save a lot of work.511 MeV. σ} is a permutation of {0. (d) Use the value of the gravitational constant GN / c5 = 6.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). Also construct and calculate the Planck length Lpl . 3} the index α can only be equal to one of the indices in ǫµνρσ . a+ . re = e2 /4πǫ0 me c2 to the Compton wavelength. c. 1.3. 0.

∂t2 (2. i ∂ ∇2 ψ(r. 2M (2. t).5) with (k 0 )2 = k2 + M 2 . namely plane waves characterized by a wave number k. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. 2 + M 2 φ(r. t) = ∂2 − ∇2 + M 2 φ(r. ∂t 2M (2. p). But the replacement 2. p2 = pµ pµ = E 2 − p2 = M 2 .1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . one obtains (2. E = p2 . 8 . (2. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite.7) k 0 = ± k2 + M 2 = ±Ek . written as pµ −→ i∂µ is covariant (the same in every frame of reference). t) = 0.1 The Klein-Gordon equation In this chapter.2) Acting on the wave function one ﬁnds for a free particle. Although it is straightforward to ﬁnd the solutions of this equation. φk (r.3) Equations 2.3 are not covariant. (2. t) = − ψ(r. (2.1 and 2.Chapter 2 Relativistic wave equations 2.4) 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.2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. ∂t p −→ pop = −i∇. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. t) = exp(−i k 0 t + i k · r).

9) They satisfy the continuity equation. 2. just as the dependence on time was in ordinary quantum mechanics.14) ˜ ˜ Introducing φ(Ek .e. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). having the KG equation as a space-time symmetric (classical) equation.10) ∂t which follows directly from the Schr¨dinger equation. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. 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. ∂ρ = −∇ · j. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. however. relativistically the density is not a scalar quantity.2) is j µ = i φ∗ ∂ µ φ or (ρ. This leads to the existence of negative densities.15) . The KG equation. k) ≡ a(k) and φ(−Ek . j) of the four-current j.13) ˜ with (in principle complex) coeﬃcients φ(k).2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. rather than as a wave function. j) = ↔ i φ∗ ∂0 φ. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. k) + ei k·x φ(−Ek .5).8) (2.11) Therefore. there are no longer fundamental objections to mix up space and time. −k) . is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. And. (2. (2. As we will see later both problems are related and have to do with the existence of particles and antiparticles.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. −i φ∗ ∇ φ . 2. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. we want the most general solution.3). 2M 2M (2. Then. for which we need the interpretation of φ itself as an operator. (2π)3 2Ek (2. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. ∂µ j µ = 0.Relativistic wave equations 9 i. This continuity equation can be written down o covariantly using the components (ρ. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . but rather the zero component of a four vector. which is what Lorentz transformations do. (2. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. ↔ ↔ (2. The appropriate current corresponding to the KG equation (see Excercise 2. there are solutions with negative energy.

which implies (xµ ) = (t. x) → (t. for which we consider space inversion. φP (x) ≡ φ(˜) (P for parity). (2π)3 2Ek (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) . one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).e. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. −x) ≡ (˜µ ).15 one has elimated k 0 and in both equations k · x = Ek t − k · x. 2. i.Relativistic wave equations 10 In Eqs 2. changing the sign of the spatial coordinates. Another symmetry is found by complex conjugating the KG equation. (2π)3 2Ek (2. This shows how parity transforms k-modes into −k modes. (2. It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.23) . The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.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) . after restricting the energies to be positive).6 We will explicitly discuss the example of a discrete symmetry. Since a · b = a · ˜ it is easy to see that ˜ b. (2. (2π)3 2Ek (2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy.14 and 2. Performing some Lorentz transformation x → x′ = Λx. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes).18) (2. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) .17) (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).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. x ∂µ ∂ µ + M 2 φ(˜) = 0. (2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k).16) The consequence of this is discussed in Exercise 2.

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

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

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

z]] + [y. In the same way we can consider rotations around the x. π) (n. x]] + [z. One way of viewing the parameter space. It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. z ) = lim N →∞ that are generated by i 0 . π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. Locally this is a 3-dimensional Euclidean space and SU (2). thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. which reﬂects itself in the noncommutation of the generators.2 π) (-n. for rotations around z for instance.7) = a0 0 −1 i 0 1 0 0 1 • [x. These generators form a threedimensional Lie algebra SO(3). Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . and thus (σ · n)2 = 1. 2 (3. [y. x]). y]] = 0 (Jacobi identity). x] = 0 (thus [x. ] that satisﬁes • ∀ x.4) R . . y] ∈ A. [z. They satisfy the commutation relations [Li . y ∈ A ⇒ [x. (3.5) Rotations in general do not commute. 0 (3.and y-axes 0 0 0 0 0 0 0 −i . n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n .8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). (3.6) Summarizing. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ).z N N or (Lk )ij = −i ǫijk . ˆ therefore. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. • [x. Next.Groups and their representations 14 (n. with real a’s and 3 (ai )2 = 1. namely that there exists a bilinear product [.1: the parameter spaces of SO(3) (left) and SU (2) (right). the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . is a 3-dimensional Lie-group. y] = −[y. Lj ] = i ǫijk Lk . ˆ R(ϕ. Ly and Lz . [x. It is straightforward to check that the conditions require a0 + i a3 +i a1 + a2 = a0 1 + i a · σ A = +i a1 − a2 a0 − i a3 1 0 + i a1 0 1 + i a2 0 −i + i a3 1 0 (3.

for ﬁxed n. . thus. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. in particular that AB → RAB = RA RB ). immediately sees that the Lie algebras are identical.11) Thus σx /2.9) The parameter-space of SU (2). 2! . In the full parameter space. one has a Lie algebra isomorphism that is linear and preserves the bilinear product.14) For a Lie group. . 2 2 2 (3. . The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . n) i ∂ϕ = ϕ=0 (3. i. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). ˆ ˆ A(ϕ. i. Near the identity. n! 2! n=0 ∞ (3. the above mapping corresponds to the trivial mapping of the Lie algebras. [g. For instance in a function space translations.e. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones.e.e. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. 3. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. i. n) −→ R(ϕ.15) [g. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. . The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. however.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. . sometimes in more than one way (but this will be discussed later).10) σ ˆ · n. also can be considered as a ﬁlled 3-sphere. 2 (3. (3. H] = 0 for g ∈ G. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). (3.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. n) ˆ −→ R(2π − ϕ.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . . .12) One. thus. (3. n) ≈ 1 + i ϕ J · n. Suppose we have a system described by a Hamiltonian H. it is suﬃcient that the generators commute with H. They satisfy σi σj σk = i ǫijk . SU (2) ≃ SO(3). H] = 0 for g ∈ G. n) π ≤ ϕ ≤ 2π.

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

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

21) g 20 g21 g22 g23 0 0 −1 0 g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as x2 = xT G x (3. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ . Writing x as a column vector and the metric tensor in matrix form.27) (3. det Λ = −1 is called improper). 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. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . e To derive some of the properties of the Lorentz group. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . e. (3.g. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. From this property.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. (ii) |Λ00 | ≥ 1. it is convenient to use a vector notation for the points in Minkowski space.25) (3. (det Λ = +1 is called proper. consisting of the Lorentz group and translations. G = 10 (3. We considered the generators and looked for representations in ﬁnite dimensional spaces. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. In this section we consider the Poincar´ group. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.26) (3.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.Groups and their representations 18 3. σ/2 in a two-dimensional (spin 1/2) case.22) (Note that xT is a row vector). 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 .

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

three of which only involve spatial coordinates (rotations) and three others involving time components (boosts).g. We therefore ﬁnd (again) that there are six generators for the Lorentz group.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . inﬁnitesimally approximated by (I + ω. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ.33) These four transformations form the Vierer group (group of Klein). (3. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . The extension to the Poincar´ group is e ↑ straightforward. we continue using covariance arguments. 3 30 with η = ω and K = M (e. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . inf (3.39) (e. Also this group can be divided into four parts. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . Summarizing. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . K = M ). [M µν .35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . P+ etc.38) Explicitly.36) (3. i i0 i i0 1 ijk M jk 2ǫ (3. ǫ).g. There are six generators. This is a + normal subgroup and the factor group is the Vierer group.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. 3. We will just require Poincar´ invariance for the generators themselves.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. Of the four parts only L↑ forms a group. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . a). explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . In order to ﬁnd the commutation relations including the translation generators. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.

The eigenvalues of these will be the four-momentum pµ of the state. P j ] = i δ ij H/c2 . ν (3. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. the generators J± in the case of the rotation group. as part of the set.44) (3. K j ] = i ǫijk K k .38 could have been obtained in the same way. the generators J 2 and J 3 in the case of the rotation group. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . P j ] = [P i . [P µ . .45) Note that the commutation relations among the generators M µν in Eq. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). K j ] = −i ǫijk J k /c2 .48) . other states in that representation are obtained by the action of operators outside the maximal commuting set. . For instance.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. [J i . [K i . a)P µ U (Λ. (3. (3. a)P µ U † (Λ. Inﬁnitesimally. (3. . . We have here reinstated c. J j ] = i ǫijk J k . P ρ ] = 0.42) (to decide on where the inverse is. These form a group called the little group associated with that four vector. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. P µ |p. = pµ |p. leads to U (Λ. They just state that M µν transforms as a tensor with two Lorentz indices. one obtains 2 [P i . P ) in terms of the Hamiltonian and the three-momentum operators. a) = Λµν P ν or U † (Λ. 3. H] = i P i . known from non-relativistic quantum mechanics are obtained. Explicitly. [J i . [J i . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. H] = [J i . the generators M αβ no longer exclusively act in Minkowski space. 2 2 (3. . P ν ] [M µν . .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. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. ǫ) = 1 − ωαβ M αβ + i ǫα P α . (3. H] = 0. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . The generators of the translations are the (momentum) operators P µ . 3.41) 2 At this point. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν .46) [K i . = −i (g µρ P ν − g νρ P µ ) . that commute e among themselves. writing the generator P = (H/c. a) = Λ−1 µ P ν . P j ] = i ǫijk P k . [K i . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . Taking any one of the states in an irreducible representation. check that U1 U2 corresponds with Λ1 Λ2 ). .47) (3. For instance.

= = i exp − ǫµνρσ M µν pρ sσ |p. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. Wν ] = −i (gµα Wν − gνα Wµ ) . i ǫµνρσ W ρ P σ . .51) The following properties follow from the fact that W µ is a four-vector by construction. p0 > 0. which can thus be chosen spacelike. 2 (3.49) (3. . 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. (3. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).56) . . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.55) commute with all generators and therefore are invariants under Poincar´ transformations.54) This will enable us to pick a suitable commuting ’spin’ operator. 0. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. . satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . M ǫijk . . Wα ] = [Wµ . 0. (3.52) (3. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . while the others have no physical signiﬁcance. . .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. . .53) (3.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . 2 exp (−isµ W µ ) |p. the third case represents the vacuum. Massive particles: p2 = M 2 > 0. Pν ] = [Wµ .

Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p).. 0. (1. show that W /M can be interpreted as the intrinsic spin. |p3 |). (3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. 0. Having made a covariant decomposition.61) one sees that S i (p) i =− W2 . M2 (3. s. W j (p)] = i M ǫijk W k (p). 0). M2 (3.59) (3.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 .58) i. .e. n1 (p). More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). 1. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 0). (3. 0.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. . 2 . (0. M J) with components W i = (M/2)ǫijk M jk . p0 > 0. In that case the vectors ni (p) are just the space directions and W = (0. 0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3.65) . 1 2 (3. (0. 3. Together with the four momentum states thus can be labeled as |M. i (3. 1.64) such that in an arbitrary frame where the four vectors p. W i (p) = −W · ni (p)). We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0.e. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 1. 0. 0). In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . ms .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). p. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0.57) (i.57 The best insight in the meaning of the operators W µ is to sit in the rest frame of the particle and put P µ = (M. 0.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. . The commutation relations [W i (p).

A massless particle. (3. W 2 and λ is derived from the general commutation relations for [Wµ . W 2 (3. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). 3. which can take any integer or half-integer value.1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. and gives [W 1 (p).68) (3. thus. Wν ]. W 2 (p)] = = = i M 2 λ(p). i W 2 (p). Exercises Exercise 3.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). λ . 1 2 24 (3. −i W 1 (p). (3.73) (3. The algebra of W 1 . however. W 1 (p)] [λ(p). The vacuum: pµ = 0. 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. Thus 2 λ(p) = J ·p . P µ .13.71) which is called the helicity. 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 . This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). |p. Note that angular momentum does not contribute to helicity as L · p = 0. This state is in physical applications nondegenerate and is denoted by |0 .70) as a commuting set with zero eigenvalues for P 2 and W 2 . 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). We don’t know of any physical relevance. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2).Groups and their representations The above does actually include the massive case. implying continuous values for the spin (actually for W i = M S i ).72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations.67) (3. is characterized by its momentum and the helicity. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. W µ (p) = λ(p) P µ . a)|0 = |0 . 3.67 has a vanishing right hand side). W 2 (p)] [λ(p). Thus we have P 2 . |p| (3. But if M = 0 one has a diﬀerent algebra (Eq. U (Λ. 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).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p).

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

. momentum and spin operators e satisfy the canonical commutation relations. Exercise 3.7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations.69 of the Pauli-Lubanski operators for massless particles. rj ] = [pi .e.. that’s why many-particle non-relativistic quantum mechanics is ’easy’. = p. i. This is possible for a single particle. p×s 1 (rH + Hr) − t p + . there is a one to one mapping) the algebra in Eqs 3. (c) Give the unitary operator for boosts.67 to 3. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. pj ] = iδ ij and [si . Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. pj ] = [ri . the others vanish. H P J K = p 2 c 2 + m2 c4 . show ﬁrst the operator identity [r. sj ] = iǫijk sk .e. i. but it can be done. although that in this case is a relation that is found already from the Ehrenfest relation for O = ri . ∂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. r2 ) inevitably leads to interaction terms in the boost operators.e. derive the commutation relations and show that the algebra is isomorphic to (i. These do not matter in the non-relativistic limit (c → ∞). boost invariance leads to a conserved quantity. Exercise 3.Groups and their representations 26 Exercise 3. the operator U (u) that gives U (u) r U −1 (u) = r − u t. Hint: for the Hamiltonian.6 Consider rotations and translations in a plane (two-dimensional Euclidean space). consistent with the (canonical) commutation relations of a quantum theory. they describe r −→ r′ = r − u t while t′ = t. [ri . [ri . (b) In quantum mechanics one has the Ehrenfest relation i dO = [O. sj ] = [pi . Comment: extending this to more particles is a highly non-trivial procedure. and U (u) p U −1 (u) = p − m u. indeed assures d K /dt = 0. Determine the commutation relations for the corresponding operators Ki in quantum mechanics. Do this by showing that the set of operators. = r × p + s. although the presence of an interaction term V (r1 . f (p)] = i ∇p f (p).5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame. (a) In classical mechanics. From either of these. H] + i dt ∂O . in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. if you don’t want to do this in general. sj ] = 0.

C) is the group of complex 2 × 2 matrices with determinant one. 2 2 σ σ J = . becoming the deﬁning representations of SL(2. j ′ ).3) (4.7) From the considerations above. Special cases are the following representations: Type I : (j. 2 2 (4. The matrices in SL(2. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. With this choice of parameter-spaces the plus and minus signs are actually relevant. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . it also follows directly that the Lorentz group is homeomorphic with the group SL(2.1 The Lorentz group and SL(2.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn.9) (4.4) (4. U (ϕ) H(η) (4. 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. K = i J (A = 0).2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). i ǫijk B k . respectively. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. They will be labeled by two angular momenta corresponding to the A and B groups. 0.5) = 1 (J + i K). B j ] = i ǫijk Ak . 2 (4. B j ] = [Ai .10) .Chapter 4 The Dirac equation 4. C).1) (4. 0) Type II : (0. K = +i . This tells us how to ﬁnd the representations of the group. 2 1 (J − i K). similarly as the homeomorphism between SO(3) and SU (2). It is simply connected and forms the covering group of L↑ . The group SL(2. K = −i . j) K = −i J (B = 0). (4.6) (4. C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ . (4. Aj ] = [B i . (j.

ǫ−1 = ǫ† = ǫT = −ǫ). U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. thus. This shows that M ∗ ≃ M . ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). The following relations for + rotations and boosts are useful. however. (4. the ﬁrst equation shows that the conjugate representation with spinors χ∗ transforming according to −σ∗ /2 is equivalent with the ordinary two-dimensional representation (±ǫχ∗ transforms according to σ/2). (4. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. while ±ǫη ∗ transforms as a type-I + spinor. U σ µ aµ U −1 = σ µ ΛR aµ .14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices.e.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. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . thus. One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. C) and.15) (4. cannot be connected by a Lorentz transformation belonging to L↑ . (4. σ µ ≡ (1.18) As already discussed for SU (2). (4. H(η) = exp(−η · σ/2). ξ −→ Hξ.16) = σ µ Λ R aµ . namely H(η) = exp(η · σ/2). ˜ = σ µ Λ B aµ . ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. −σ). i. In fact. C). The Lorentz transformations Λ(M ) and Λ(M ). (ǫ∗ = ǫ.19) . Λ(ǫ) ∈ L+ . there exists the matrix ǫ = iσ 2 . ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. ˜ (4. (4.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. such that M = SM S −1 . if there does not exist a unitary matrix S. C) or L↑ . H ≡ (H † )−1 = H −1 ).11) Considering ξ and η as spin states in the rest-frame. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. This is not true for the two-component spinors in SL(2. C).12) ξ → U ξ.1). η → U η.13) 2M (E + M ) (exercise 4. Eqs 4. σ). one can use a boost to the frame with momentum ˆ ˆ p. 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. This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . η → Hη.

13. therefore.23) u(p. the boost in Eq.17. (4. −→ −r (vector). Taking M (Λ) = H(p). The i angular momentum operators J are represented by σ i /2. C). and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. (pseudoscalar). (vector). The behavior is the same for classical quantities.e. we obtain for the two components of u which we will refer to as chiral right and chiral left components. etc. a) and ˜ a = (a0 . We also have seen that the representations (0. The representing member of the class P− is the parity or space inversion operator Is . that have the correct commutation relations.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. = (4.g. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . −→ M (Λ) (4. i. −a). 0). they can be connected. 0) of SL(2. 4. m) = χm uL H(p) 0 .21) In nature parity turns (often) out to be a good quantum number for elementary particle states. Within the Lorentz group. 1 ) and ( 1 . (axial vector). Although these operators cannot be used to label the representations. they cannot be connected by a unitary 2 2 transformation. 4. Thus M ∗ and M are not equivalent within SL(2. must combine the repe resentations. we have seen that in the rest frame W i (p)/M → J i . It has the same eﬀect as lowering the indices. but to L↑ .1. but by a transformation belonging ↑ ↑ to the class P− .The Dirac equation 29 where The matrix I2 does not belong to L↑ . 0) are inequivalent. A parity transformation changes aµ into aµ . From the deﬁnition of the representations (0. 0 χm uR H(p) . m) = η(0. as we have seen in section 4. 2 ) and ( 2 .20) 4. 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. ǫµνρσ will change sign. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . 1 ) and ( 1 . i. the well-known two-component spinor for a spin 1/2 particle.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. (scalar). e. C) are suitable for representing spin 1/2 particles. generators. where a = (a0 . −→ −p (vector).e. the aij elements of a tensor will not change sign. but rather the operators W i (p) should be used. ) −→ ( . m) = χm . 2 2 (4. For the spin 1/2 representations of the Poincar´ group including parity we. For a particle at rest only angular momentum is important and we can choose ξ(0. This provides the easiest way of seeing ˜ what is happening.

. p (4. (4. γ ν } = 2 g µν .29) which is an explicit realization of the (momentum space) Dirac equation. 4. {γ µ . which in general is a linear equation in pµ .7. We will discuss another explicit realization of this algebra in the next section. (4. The ﬁrst indeed is solved.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 .31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. (4. 0 H 2 (p) uR uR .27 is known as the Weyl representation.32) suppressing on the RHS the identity matrix in Dirac space. (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. It is of a form that we also played with in Exercise 2. = H −1 (p). The general requirements for the γ matrices are thus easily obtained. = (4. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately).30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. 2 (4. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. γ } ∂µ ∂ν + M 2 ψ(x) = 0. (iγ µ ∂µ − M ) ψ(x) = 0. Since ∂µ ∂ν is symmetric.33) deﬁne for a four vector a the contraction a = aµ γµ / . As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. Applying (iγ µ ∂µ + M ) from the left gives (4. (4.24) (4. The explicit realization appearing in Eq.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). From this one obtains the Cliﬀord algebra for the Dirac matrices. this can be rewritten 1 µ ν {γ .28) which is a covariant (linear) ﬁrst order diﬀerential equation.

From the vacuum a particle can be removed. . Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. (4. (4. but with properties such that it can be annihilated by the particle (e.37) Using the explicit form of γ in the Weyl representation (see Eq.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. The Dirac sea forms the vacuum. 4. This hole forms an antiparticle with the same mass. .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 . remains. The second problem encountered before. the one of the negative energy states. γ ν } = 2 g µν . opposite charge).The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4.35) is indeed positive and j 0 can serve as a probability density. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.36) (4.3 General representations of γ matrices and Dirac spinors {γ µ . E = +M (twice) and E = −M (twice). This problem was solved by Dirac through the introduction of a negative energy sea. Dirac supposed that the negative energy states were already completely ﬁlled and the exclusion principle prevents any more particles to enter the sea of negative energy states. (4.g.27).34) for the adjoint spinor ψ ≡ ψ † γ0 . 4. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). The density j 0 = ψγ 0 ψ = ψ † ψ (4. This can most easily be seen in the particle rest frame. Relying on the Pauli exclusion principle for spin 1/2 particles.1: The Dirac sea of negative energy states and antiparticles. one sees that there are two positive and two negative eigenvalues. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.

R. γ σ ]. −→ Sψ. which in the case of the Weyl representation are just the upper and lower components.3) . = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . γ 2 ].43) S=√ . =ρ ⊗1= 0 −1 3 (4. that project out chiral right/left states.45) (γ5 )W. = S(γµ )W.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . 2 (4.R. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . 0 (4. −→ Sγµ S −1 . γ 0 ]. 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . (4.46) (4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ.44) and an adjoint spinor deﬁned by ψ = ψ † γ0 .R. (γ5 )S.41) (4. We note that ψ → Lψ and ψ → ψ L−1 .9 and 4. γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices. 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 . while L−1 γ µ L = Λµ γ ν .R. 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .The Dirac equation 32 The Weyl (or chiral) representation: 0 1 .47) (4. 4! (4. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. S −1 .10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .40) (4. For instance in the Weyl representation (but valid generally).48) are projection operators. (γµ )S. is 1 1 1 (4. the rotation and boost generators in Weyl representation in Eqs 4. γ µ } = 0 and explicitly one has 0 1 1 . For example.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.

usually to be taken unity.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. (C)W. ∂ (4.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.59) −→ ψ c = C ψ = η ǫ ξ∗ . ψR . = = 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.56) (i/ − M ) ψ c (x) = 0.55) (4. ∂ (4. for instance. called charge conjugation. x) → x = (t.6). (C)S. parity. ψ= (4. This symmetry does not change the spin 1/2 nature. One has ψ c = −ψ T C −1 . T (4. such that T Cγµ C −1 = −γµ . In S.58) where ηc is an arbitrary (unobservable) phase. In any representation a matrix C exist.R. is a left-handed spinor. As with parity we look for a transformation. which is again a solution of the Dirac equation. x = (t. Some properties of C are C −1 = C † and C T = −C.g.g. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.53) (4.) and all representations connected via a real (up to an overall phase) matrix S.R. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . reverse the charge of the particle. but it does. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . that brings ψ → ψ c .R. ψ= (4. e.R.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. T (4. 0 −ǫ 0 0 −ǫ 0 . = 0 ǫ iσ 2 (4. C is real and one has C −1 = C † = C T = −C c and [C. −iσ 2 0 −ǫ = . 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. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. (i/ − M ) ψ(x) = 0. ∂ with the solution ψ c (x) = ηc Cψ (x). The latter implies that the conjugate of a right-handed spinor.52) η ξ one can apply space inversion. γ5 ] = 0.57) (4. Explicitly. (or W.54) e.

v (p.70) = Ps = ˆ 0 1∓σ·s 2 2 . 2M 2M (4. s)v(p.63) u(p. s) u = . The ¯ ¯ solutions are normalized to u(p. After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) = s d3 k u(k. p (4. ⇔ (/ − M ) u(p) = 0. s ) = v (p. be obtained by a Lorentz transformation. (2π)3 2Ek (4. In that frame is a spacelike unit vector sµ = (0.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s)v(p.62) where χs are two independent (s = ±) two-component spinors.68) (4.The Dirac equation 34 4. s) e−i k·x b(k. s). 2M 2M −/ + M p v(p.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). s′ ) = 2Ep δss′ . s) = v(p. s) and C v T (p. s′ ) = 2M δss′ .. s)u(p. with E = Ep = + p2 + M 2 . s)u(p. s ) = v † (p. the spinors satisfy C uT (p.1). In an arbitrary frame one has s · p = 0 and s(p) can e. in which γ 0 is diagonal (see discussion of negative energy states in section 4. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. the spin polarization vector in the rest frame is the starting ˆ point. ψ(x) = v ±s (p) ei p·x .4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. s ) = 0. ∂ −iσ · ∇ −i ∂t − M (4. s). p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. ⇔ (/ + M ) v(p) = 0. s) = Ep + M σ ·p Ep +M χs . ¯ ¯ † ′ u (p. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. s) + v(k. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. s′ ) = −2M δss′ . v(p. s)v(p. s) = u(p.69) = − s In order to project out spin states. s) v = . (4. s) . ¯ (4. ψ(x) = u±s (p) e−i p·x .60) ψ(x) = 0.64) (4.61) σ·p −(Ep + M ) There are also two negative energy solutions. ¯ ′ ′ Explicit solutions in the standard representation are χs .65) (4. s) = u(p.g. s)u(p. where u satisﬁes −σ · p Ep − M u(p) = 0. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s) ei k·x d∗ (k. the best representation to describe particles at rest is the standard representation. s)¯(p. s)¯(p.

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

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

where x = (t. a 4 a · b. Exercise 4. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). Show that the equality holds. 4 2 Exercise 4. by letting them act on the four basis spinors u(p. x ˜ . 4. e. s)¯(p. x → x.1 Prove Eq. 1 i −i S ρσ = − [γ ρ . They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . 0 if n is odd.g. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. γ σ ] = σ ρσ .3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν .77) Exercises Excercise 4.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. ˜ Exercise 4.13. s) u p /+M = 2M 2M both sides are projection operators. H(p) = exp η·σ 2 = M +E+σ·p . (4. 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.g. of which only the traces of four and two gamma-matrices are nonzero. s). Do this e. −x) is also a solution of the Dirac equation. Excercise 4.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices.6 Apply space-inversion. s) and v(p.5 Show starting from the relation {γµ .

Calculate for now just |M |2 .511 MeV).66 MeV. the decay would even be forbidden.The Dirac equation 38 Exercise 4. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. mµ = 105. 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. The pion has spin 0. . We will not pursue this chiral freedom at this point and take α = 0. namely i / ψ = eiαγ5 ψ. which is the most important factor needed to ﬁnd the decay width (inverse lifetime).8 × 10−6 . |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. r). i. there is no current ﬂowing through the surface of the cavity. you need to determine the corresponding conﬁnement condition for ψ. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. Calculate the value of k for M R = 0. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. (b) Next. Note: To prove this. 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. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. n It turns out that there is actually a whole family of conditions that provide conﬁnement. Do you understand why? The calculation of the life-time is completed in chapter 11.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x.e. M R = 1. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. me = 0. we want to conﬁne this solution to a spherical cavity with radius R. For such a cavity nµ = (0. ν p Note: how such parametrizations work will be explained in chapter 8.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). Excercise 4. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. Our starting point is the transition amplitude given by M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ). 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. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal).5 and M → ∞.57 MeV.

1 Fields for spin 1 In section 3. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. ǫλ ). with ∂µ V µ = 0 (Lorentz condition).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. just as the two-component Dirac spinors did for spin 1/2. (2 + M 2 )Vµ = 0. These can describe spin 1 states in the rest-frame.Chapter 5 Vector ﬁelds and Maxwell equations 5. 39 (5. M2 (5.6) where Aµ is the four-vector potential (5. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. In an arbitrary frame ǫµ (k) with λ = 0. λ) + ǫ(λ)∗ (k) eik·x c∗ (k. These are solutions of the Klein-Gordon equation with in addition a condition.5) −E B2 .7) .2) (5. λ) . A). where k = k0 = (M.1) The Lorentz condition implies that k µ ǫµ (k) = 0. The vectorﬁeld is therefore (λ) suited to describe spin 1.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. −B 1 0 3 With the electromagnetic ﬁeld tensor (5.4) λ=0. (5.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. namely ǫµ (k0 ) = (0. In the restframe.± 5. ±(1) are obtained by boosting the restframe vectors. µ (2π)3 2E µ (5. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . there are then indeed three independent possibilities. 0).

5. As remarked above. |k|). while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). in accordance with what we found in section 3. (5.8) (5. (5.17) (radiation. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. moree over.13) Hence. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). 0). for M = 0. One possible choice is the gauge A0 = 0 (5. ∂µ Aµ = 0.11) This equation is not aﬀected by a gauge transformation. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . Furthermore. It states the absence of magnetic charges and Faraday’s law. The equation in vacuum. ∓1.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. (5. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . Although the Lorenz condition is a constraint. For electrodynamics one has furthermore the freedom of gauge transformations. transverse or Coulomb gauge).e. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . . if ∂µ Aµ = 0 for large times |t|.15) (5. if in Eq. Therefore if k µ = (|k|. Taking the divergence of this equiation.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. (5. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. AP (x) = Aµ (˜). 40 (5.15 the d’Alembertian is replaced by 2 + M 2 . only two independent vectors remain ǫ(±) µ (k) = (0. i. 0. j) the Maxwell equations read ∂µ F µν = j ν .12) (5.5. (5. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. 2Aµ = 0.16) (5. it will vanish at all times. in which case one has 2Aµ = jµ . −i. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. ˜ x Note that via parity transformation one obtains another solution. 0.10) the electric and magnetic ﬁelds are unchanged.14) of which the solutions give the Li´nard-Wiechert potentials. Under a gauge transformation Aµ −→ Aµ + ∂µ χ.

Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. 0). The situation. illustrated in ﬁg. To be precise.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. nevertheless. sin ϕ. λ L − λ λ or ∆x = (L− λ/d)δ. 2 2 2 BR2 BR2 y BR2 x A= . but A = 0 (hence there is structure in the vacuum). . The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . It is however. = 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. may look a bit akward. ϕ= − ˆ 2r 2r2 2r2 where r = (cos 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. (5. 5. F = e E + ev × B (5. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). but where E = B = 0.0 . 0) and ϕ = (− sin ϕ. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). This phase diﬀerence is actually observed. Inside solenoid: Outside solenoid: B = B z.0 . The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). This occurs in the region outside the solenoid. This causes an interference pattern as a function of ∆x.19) are gauge invariant. . where Aµ = 0.1.1: The Aharonov-Bohm experiment 5. A= By Bx Br ϕ= − ˆ .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. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. . while the electromagnetic ﬁeld Aµ is not. cos ϕ.

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

∂t ∂B = 0. . ∇×B =j+ ∇×E+ ∂E . 5.8 and 5.1 Show that Eqs. ∂t ∇ · B = 0.9 explicitly give the equations ∇ · E = ρ.Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5.

x) ˙ (6. x (t) = x(t) + δx(t). x) = K − V = ˙ 1 mx2 − V (x). recall classical mechanics with the action t2 S= t1 dt L(x.6) The ﬁrst term leads to the Euler-Lagrange equations. and the total change x′ (t′ ) = x(t) + ∆x(t).7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. ∂x ˙ (6. thus t′ = t + δτ. d dt ∂L ∂x ˙ = ∂L . (6. ˙ p= ∂L .4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 . ˙ 2 (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time.5) with ∆x(t) = δx(t) + x(t) δτ .3) (6. As an example.1) and as an example the lagrangian L(x.8) 44 . These equations can be obtained from a lagrangian using the action principle. ∂x (6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. t1 (6.Chapter 6 Classical lagrangian ﬁeld theory 6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation).

. t1 ) and R3 . their derivatives and possibly on the position. x) ≡ p x − L . t1 (6.14) (6. t2 ). In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds. = φ(x) + δφ(x) (6. . ∂µ φ. (R3 . x′ ) − d4 x L (φ.11) λ=0 of which p and H are the simplest examples related to space and time translations. Consider a lagrangian density L which depends on these functions. λ=0 (6. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). δ(∂µ φ) (6. = ∂R (R3 .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ).9) which is done because the ﬁrst term (multiplying ∆x. which in classical mechanics vanishes at the boundary) does not play a role. dσµ . but which in 3 dimensions includes conserved quantities related to rotations and boosts. L (φ(x).t1 ) (6. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ . indicated with δF [φ]/δφ should pose no problems.6 can be rewritten as δS = . − d3 x . + ∂L ∆x − H δτ ∂x ˙ t2 . x).13) (6. ˙ (6.t2 ) d3 x . ∂µ φ(x)). x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). The hamiltonian H is deﬁned by H(p.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. ∂µ φ(x)) = R d4 x L (φ(x). . ∂µ φ(x).Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. Variations in the action can come from the coordinates or the ﬁelds. . We will come back to it in a bit more formal way in section 9.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ.18) 1 Taking a functional derivative. 6. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . In general any continuous transformation.12) Here R indicates a space-time volume bounded by (R3 . (6. ∂µ φ′ . . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). ˙ x The second term in Eq. . .e. with ∆φ(x) = δφ(x) + (∂µ φ)δx . ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . .2. H is a conserved quantity. i.15) d4 x′ L (φ′ . R (6. . .

For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities.27) (6.23) (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. leading to the Euler-Lagrange equations. 2 ψ = 0.19) δ(∂µ φ) δφ 6.20) (6. 2 (6.26) The Dirac ﬁeld The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ.29) .Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration). 2 ∗ (6. (6. One easily obtains (2 + M 2 )φ(x) = 0. δL δL ∂µ = .24) (6. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0).2 Lagrangians for spin 0. (6. (6. The integrand of the ﬁrst term must vanish. leading to (2 + M 2 )φ(x) = 0. Using the variations in ψ (in the symmetric form). (2 + M )φ (x) = 0. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found.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. ∂ (6.21) which diﬀer only by surface terms.25) (6. the second term is irrelevant.

. ∂ ∂ 2 2 (6. (6. and thus (β ∗ α)∗ = α∗ β = −βα∗ . Vector ﬁeld From the lagrangian density L 1 1 λ 2 = − Fµν F µν + M 2 Vµ V µ − (∂µ V µ ) 4 2 2 1 µ = V (∂ 2 + M 2 )gµν − (1 − λ)∂µ ∂ν V ν . (6.g. 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. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . Using the twospinors ξ and η. that the lagrangian separates into two independent parts for M = 0.34) (6.36) (6.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. 1 M η † ǫη ∗ − M ∗ η T ǫη . the mass term in the Dirac lagrangian 6. which comes because we in essence use ’real’ spinors. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR .31 is given by LM (Dirac) = −M ξ † η + η † ξ .and lefthanded ones.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).32) There exists another possibility to write down a mass term with only one kind of ﬁelds. ψL ≡ (ψL )c = C ψL = 0 (6. or equivalently separate the ﬁeld into right. 2 αβ = −βα.31) showing e.g. (6.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. but note the factor 1/2 as compared to the Dirac lagrangian. The reasons for Grassmann variables will become clear in the next chapter. The Majorana case is in fact more general. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 .37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . namely2 LM (Majorana) = + (6.7).38) (6. 2 (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left ﬁelds ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors. Peshkin and Schroeder or Exercise 12. We note that of which the left part coincides .

These do not aﬀect the dynamics and will give rise to conserved currents. . these conserved quantities become operators. δ(∂µ φ) For continuous transformations. which in a consistent picture precisely generate the symmetries.44) δL ∆φ + Θµν (x)δxν . consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. (6. 6.. . t). 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. one can consider the variations at the initial or ﬁnal surface.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . Returning to δS the surface term is rewritten to δS = R d4 x . As an example for the classical ﬁeld.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). requires the presence of a conserved quantity. discussed in the next chapter. t). (6. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. (6.46) 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 .3 Symmetries and conserved (Noether) currents Next.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . In particular when the lagrangian is invariant under symmetries. + ∂R dσµ dσµ ∂R ≡ where d x . In this case. (6. we consider the second term in Eq. Q= d3 x J 0 (x.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .40) 6.47) .σ1 with σ = (R3 . the presence of a symmetry that leaves the lagrangian invariant. Q(t1 ) = Q(t2 ).41) (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν .. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . which is the space-integral over the zero-component of a conserved current. (6. δ(∂µ φ) (6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). δ(∂µ φ) (6.18. ∂µ J µ (x) = 0.

50) (6.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). we obtain (since δxµ = 0).51) (6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .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. ˙ Q = d3 x j 0 (x.53) These are the energy and momentum.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. From Noether’s theorem one sees that the current Θµν ǫν is conserved. (6. (6. 2 (6.49) These currents are conserved as discussed already in chapter 2.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.g. H = d3 x Θ00 (x). Summarizing. omitting the parameter Λ jµ = − δL δL (i e ψi ) (−i e ψi ) − δ(∂µ ψi ) δ(∂µ ψ i ) 49 = e ψγ µ ψ. (6.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. e.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.54) (6.3) j µ = i e φ∗ ∂µ φ. including e the space-time translations and the Lorentz transformations. 1 (6. (6. The integral over the zero-component. ↔ (6. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . = 0. t) is the conserved charge (Q = 0). = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). In the next section we will see the quantity show up as the charge operator. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . 6. .

φ(x). ∂µ T µν (6.59) (6. (6. In general this tensor is not symmetric. (6. 6. one has a tensor that satisﬁes (exercise 6.58) (6. however. independent of x. Deﬁning T µν = Θµν − ∂ρ Gρµν .Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. where H ρµν is the quantity appearing in M ρµν . In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4.70) .5 (Abelian) gauge theories φ(x) → ei eΛ φ(x). e. is not invariant under local gauge transformations. (6. the angle of the transformation depends on the space-time point x. • Vector ﬁeld Aµ : −i Sρσ = aρσ .68) (6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.65) M µρσ = T µρ xσ − T µσ xρ . φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).g. (6. in which the U (1) transformation involves an angle e Λ. (6.61) A ﬁnal note concernes the symmetry properties of Θµν .66) In section 6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ .3 we have seen global transformations or gauge transformations of the ﬁrst kind. 2 (6.64) (6. Any lagrangian containing derivatives.63) µν = ∂µ Θ .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 .69) i eΛ(x) φ (x).6) T µν = T νµ . Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).67) i.e.3). ∂µ φ(x) + i e ∂µ Λ(x) e (6. −i eΛ(x) ∗ i eΛ(x) (6.

The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. ∂ 2 4 (6.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. 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).73) (6. 4 (6. For this purpose it is necessary to introduce a vector ﬁeld Aµ . 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. Dµ φ) − Fµν F µν . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. (6. As an example consider the Dirac lagrangian.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 .71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.78) . 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. (6. (6. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). 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 . 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν . −e j µ Aµ = −e ρφ + e j · A. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld.72) (6. D ∂ A (6. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. ∂µ φ) =⇒ L (φ.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).

1 (a) Show that the Klein-Gordon equation for the real scalar ﬁeld can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. φ → eiΛ φ. ∂µ F µν = j ν . ∂µ F µν = j ν . . (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.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .e.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.79) Exercises Exercise 6. i. where j µ = e ψγ µ ψ. −eψγ ν ψ. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . and deduce what is the charge of the antiparticle as compared to a particle. give the equation which is satisﬁed by ψ c = Cψ .4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. 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. ∂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. giving the Maxwell equation coupling to the electromagnetic current. Exercise 6.

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

Quantization of the theory is achieved by imposing canonical commutation relations between q and p. q(q. labeled by the position. there are conserved ˙ ˙ quantities Q (Eq. ˙ L(q. dλ Examples are dO = [O. q) also considered in the previous chapter. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. Q]P . 6. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. (7. Further.11) found through Noether’s theorem. t). p]P . however. namely [q. in our example p = mq.5) ∆ x ∆3 x 54 . starting from the lagrangian L(q. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. (7. p]P = 1.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. dq dp dq dp (7. p] = i. is the Poisson bracket [A.3) For variations δO(p. [q. 1 qx (t) = 3 d3 x φ(x. H]P dt and dO = [O. The bilinear product for the conserved quantities. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. An immediate generalization for ﬁelds can be obtained by considering them as coordinates.Chapter 7 Quantization of ﬁelds 7.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t). ˙ 2 (7. B]P = dB dA dA dB − .2) H(p. (7. q) = ˙ 1 mq 2 − V (q).4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. q) = p q − L(q.

˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ).7) d3 x L (φ(x). Φ(f ) = d4 x φ(x) f (x). ∂µ φ(x)) = ˙ d3 x L (φ.14) In fact. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x).12) (7. t)] = [Π(x. albeit with ’sharp’ functions. t). t)] = 0. a) φi (x)U (Λ. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . qx+∆x ) . qx .2. 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. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. discussed in the section 6. t). 2 2 2 (7.11) (7. a) = Rj (Λ−1 ) φj (Λ−1 x − a). φ(y. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . Sometimes it is useful to keep in mind that. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x.8) (7. The essential conditions to consistently quantize a theory then are the following. the discretization procedure above is an explicit example. • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f .16) . (7. ∇φ) (7.6) ∆3 x L x (qx .15) with furthermore the relations [φ(x.Quantization of ﬁelds 55 etc. φ. Π(y.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. (7. t)] = i δ 3 (x − y). formally. (7. ∂ qx ˙ ∂ qx ˙ (7. ∆φ. for symmetry transformations.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. for the scalar ﬁeld theory. Π(y. i U † (Λ. t). The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. • Poincar´ invariance e The full variation of the ﬁelds. As an example. ˙ δ φ(x) (7.

[Q. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . The measurements cannot inﬂuence each other or [Φ(f ). a] = [a† .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.18) with S µν given in 6. j (7. P ] = i.24) It is straightforward to ﬁnd the commutation relations between N and a and a† . 2 i ωµν (S µν )i .23) The hamiltonian in this case can be expressed in the number operator N = a† a. a† ] = 0. 2 2 (7. (7. φ(y)] = 0 if (x − y)2 < 0.19) 7. (7. Mµν ] = i i∂µ φi (x).4. (7. Φ(g)] = 0. a] = −a. ω a† a + 2 2 ω i − [Q. Pµ ] = [φ (x). sometimes referred to as second quantization.Quantization of ﬁelds 56 or if U (Λ.25) . 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. P ] = 0 (7. j 2 one has [φi (x).22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). Q] = [P. a† ] = a† . a† ] = 1. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. and [N.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . P ] 2 (7. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. [a.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. [N.17) (Sµν )i φj (x). [φ(x). H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . Microscopic causality implies local commutativity.

4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). (2π)3 2Ek (7. i. which represents the ’number of particles’ with momentum k. a† (k ′ )] = 0. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators.26) a(k) e−i k·x + a† (k) ei k·x .32) (7. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.like’ commutation relations between a(k) and a† (k ′ ). The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . e. † √ a and a act as raising and lowering operators.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. but one needs to realize that in that case k 0 = Ek = k2 + M 2 .29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator .31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2.e. (n − 1) a|n † i.g. Note that we will often write a(k) or a† (k).33) . a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). is equivalent with [a(k). and we see that a state |0 must exist for which N |0 = a|0 = 0. the real scalar ﬁeld φ(x) becomes d3 k (7. [φ(x). a(k ′ )] = [a† (k). Π(x )]x0 =x′0 = 0. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . (7. 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. and (7. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n .30) [φ(x).28) (7. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . 7. φ(x )]x0 =x′0 = [Π(x).e. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.

42) (7. is solved by subtracting it as an (inﬁnite) constant. This term will be adressed below. (2π)3 2Ek (7. i. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). (2π)3 2Ek (7.40) This procedure is known as normal ordering. . and multiparticle states |(k1 )n1 (k2 )n2 . d3 k a(k) e−i k·x . (2π)3 2Ek d3 k a† (k) ei k·x . 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).35) where the vacuum contribution disappears because of rotational symmetry.34) where V = (2π)3 δ 3 (0) is the space-volume. (2π)3 2Ek (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). For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. writing all annihilation operators to the right of the creation operators.46) (7. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek .45) (7. The rest of the states are then obtained from the groundstate via the creation operator. . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7.Quantization of ﬁelds 58 where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy. 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. deﬁning particle states |k = |k (with positive energy. . (2π)3 2Ek (7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k).37) (7. |k = a† (k)|0 .41) (7. . (2π)3 (7. d3 k |k k|.39) The problem with the vacuum or zero-point energy. . which now contains an inﬁnite number of oscillators.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ).e. a(k)|0 = 0. k 0 = Ek = k2 + M 2 ). |0 . n1 ! n2 ! (7.44) p|φ− (x)|0 = ei p·x .

i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . that they satisfy the required commutation relations in Eqs (7. The last item to be checked for the scalar ﬁeld are the causality condition.50) or as integrals over d4 k.51) (7. In order to calculate [φ(x).52) The result for the invariant commutator function is [φ(x). (2π)4 (7. x) = 0 and hence ∆(x) = 0 for x2 < 0. i. φ(y)] consider µν [φ+ (x). (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x .56) = −δ 3 (x). = − (2π)3 2Ek (7.49) (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. 2π (7. a† (k ′ )] 3 2E ′ (2π) k [φ− (x).18). d3 k ei k·(x−y) = −i∆+ (y − x). 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 ).54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .e. (2π)3 2Ek ≡ i∆− (x − y). φ− (y)] ≡ i∆+ (x − y). d3 k = (2π)3 2Ek = (7. φ+ (y)] d3 k e−i k·(x−y) . ∆(x) = − ǫ(x0 ) δ(x2 ).48) (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.17) and (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. (iii) ∆(0.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). (7.Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group.53) t=0 .

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) . Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). φ† (y)] = i∆(x − y).59) (7.60) = i φ∗ ∂µ φ. (2π)3 2Ek (7. (2π)3 2Ek (7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). 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 . 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) . = ∂{µ φ∗ ∂ν} φ − L gµν . a† (k ′ )] = [b(k).e. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). we will consider it as a repetition of the quantization procedure.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld.65) 7.Quantization of ﬁelds 60 7. (7. (2π)3 2Ek (7.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x).57) (7.63) i.66) . b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. extending it with the charge operator and the introduction of particle and antiparticle operators.61) (7. 2 From the lagrangian density (7.62) (7.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ. ↔ (7. which is equivalent to the relations (only nonzero ones) [a(k).

s)e−i k·x + d† (k. (2π)3 2Ek (7. s)v(k.75) (7. interchanging creation and annihilation operators.78) (7. u v (2π)3 2Ek (7. s)ei k·x + d(k. s)¯(k. s). s)e−i k·x .68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). ↔ i ψγµ ∂ν ψ − 2 (7. . The quantized ﬁelds are written ψ(x) = s d3 k b(k. s) e−i k·(x−y) j s † + vi (k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . (2π)3 2Ek (7.71) (7. s)ei k·x .e. s)u† (k.77) Note that achieving normal ordering. ∂ 2 ↔ (7. The solution is the introduction of anticommutation relations. † {ψi (x). jµ Θµν = = ψγµ ψ. s)vj (k. then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . s′ )} = {d(k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.67) i ψ / ψ − M ψψ gµν . b† (k ′ .73) (7. s)u(k.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. s) ei k·(x−y) x0 =y 0 .Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. d† (k ′ . (2π)3 2Ek d3 k b† (k.74) (7. (7.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . s).69) (7. {b(k.70) where the last line is obtained by using the Dirac equation. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. (7. s)¯(k.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. i.

we would have obtained [ψi (x). as the vanishing of Π0 induces a constraint. −λ(∂ρ Aρ ).Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. ψ j (y)] = (i/x + M )ij i∆1 (x − y). but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. ˙ δ Ai (7. It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .88) .85) which reﬂects the gauge freedom.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. For the scalar combination {ψi (x).87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.6 The electromagnetic ﬁeld 1 L = − Fµν F µν . (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 Π0 Πi δL = 0. ψ j (y)} at arbitrary times one has {ψi (x). one has † {ψi (x). ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y). ∂ (7. but has the problem of being noncovariant. ∂ (7. ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7.80) 3 = δ 3 (x − y) δij . 7.84) (7. ˙ δ A0 δL = F i0 = E i .83) From the lagrangian density the canonical momenta are = = (7.82) where i∆1 = i∆+ −i∆− . When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. (7.86) (7. 4 2 This gives the equations of motion discussed before.

0)c(k. λ)ei k·x .e. (7. ˙ ˙ . a longitudinal photon and two transverse photons. where Q is the conserved quantity in Eq. for which it is suﬃcient that ∂µ Aµ |A = 0. q) that dO/dλ = [O. λ)|A = 0. containing a time-like photon. j = 1. 7. Q]P . one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.93) which does exhibit problems with the time-like photon. i. 0) . It gives 3 k µ ǫ(λ) (k)c(k. (7. µ λ=0 (7. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. λ) − c† (k. Πν (y)] = 0. and [ℓi . and are equivalent with [c(k.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). [ℓi . rj ]P .Quantization of ﬁelds 63 for physical states |A and |B .95) Exercises Exercise 7. 2.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. which reproduce the Hamilton equations. pj ]P . λ)c(k. 6. In fact the commutation relations imply ˙ [Aµ (x). Πν (y)]x0 =y0 = i gµν δ 3 (x − y). The canonical equal time commutation relations are [Aµ (x). These problems are solved by the Lorentz constraint between physical states given above. (7. 0. λ)e−i k·x + c† (k. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k. (b) Proof for a quantity O(p. pj ]P .11. Aν (y)] = [Πµ (x).3) satisfy the conditions of a Lie algebra. λ).92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). ′ (7.94) Choosing k µ = (|k 3 |. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. c† (k ′ . 0. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. i. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k.e. what are [ri . (d) Use Poisson brackets to calculate q and p.91) (7.89) with four independent vectors ǫµ . µ λ=0 (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). Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. e−i a requires [φ(x). (2 + M 2 ) ∆inhom (x) = −δ 4 (x).2 Prove that the equal time commutation relations between quantum ﬁelds φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ ) Exercise 7. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). (b) Check the above commutation relation [φ(x). we can write for ∆R an integral where the k 0 integration remains along the real axis. Pµ ] = i∂µ φ(x). (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. ∆hom (Λx) = ∆hom (x) if f is an invariant function.Quantization of ﬁelds 64 Exercise 7. Give also the expression for ∆C .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. (2 + M 2 ) ∆hom (x) = 0.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. (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. (e) Show that ∆R (x) = 0 if x0 < 0. Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators. f (Λk) = f (k) for Lorentz transformations Λ. It is invariant.

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

x (8.1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x). and C quantity t r xµ E p pµ L s ˆ λ = s·p P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ T -t r −˜µ x E -p pµ ˜ -L -s λ C t r xµ E p pµ L s λ 66 .Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. (iγ µ ∂µ − M ) ψ(x) = 0. time reversal (T) and charge conjugation (C). writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x).1: The behavior of classical quantities under P.1 Parity xµ = (t. ˜ (8. starting with the Dirac equation for ψ(x). T. parity (P). Table 8. Both ψ p and ψ satisfy the Dirac equation. 8. T and C for classical quantities is shown in the table 1.2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space. We can determine A. −r). The consequences of P. r) −→ xµ ≡ xµ = (t.

λ)v(k. λ) Pop = ηP b(k. λ)u(k. λ)γ0 v(k. ˜ Pop d(k.11) (8.4) (8. λ)γ0 u(k. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). op P (8. λ)v(k.5) (8. if helicity λ is used instead of the z-component of the spin m. the above operation reverses the sign of λ). −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.7) = λ d k x x ηP b(k. . x (8. −λ)u(k. −λ)v(k. iγ µ† ∂µ − M ψ(˜) = 0. ˜ −v(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. −λ).3) (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. one can also consider the scalar ﬁeld and vector ﬁelds. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. m) γ0 v(k. 8.6) (check this for the standard representation.e. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. m).13) ψ(x) −→ ψ c (x) = ηC C ψ (x). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. λ)u(k. λ) e−ik·x − d† (k.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. λ) P −1 = −η ∗ d(k.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8.9) (8. λ) e−ik·˜ + d† (k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. −λ). γ x (8. (2π)3 2Ek (8. −λ) e−ik·x − d† (k. −λ) e−ik·x − d† (k. x The latter behavior of the vector ﬁeld will be discussed further below.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). m). λ) eik·x .10) i. (i˜ µ ∂µ − M ) ψ(˜) = 0. In the same way as the Fermion ﬁeld. m) = = ˜ u(k.

x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. λ) eik·x . . λ) −1 Cop = ∗ ηC b(k.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . x x (8. m). we start with the complex conjugated x Dirac equation for ψ. λ)v(k. r). it is antilinear1 . λ)C v T (k.e. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. (8. i.3 Time reversal xµ = (t. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0.19) Cop d(k. x iγ µT ∂µ − M ψ(−˜) = 0. u(k. λ). λ) Cop = ηC d(k. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜).21) where A is a 4 × 4 matrix acting in the spinor space. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. x (8. (2π)3 2Ek (8.14) (8. x (8. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ .15) (where one must be aware of the choice of spinors made in the expansion. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. The same relations hold for helicity states.16) d k ηC d(k. Top |φ = φt | = (|φt )∗ .18) (8. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . λ)u(k.Discrete symmetries 68 the latter being also a solution of the Dirac equation. m) ¯ = = v(k. λ)C uT (k. λ) e−ik·x + b† (k. r) −→ −˜µ ≡ −xµ = (−t. λ) e−ik·x + b† (k. λ). m). ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. (8. The (time-reversed) Dirac equation becomes. As time reversal will transform ’bra’ into ’ket’. as discussed in section 4). 8. antilinear operator is anti-unitary if A† = A−1 . m) ¯ C v T (k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.17) 3 = λ This shows that −1 Cop b(k. Norm conservation requires Top to be anti-unitary2 .

4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. there are 16 independent of these bilinear combinations. op T (8. −p. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. Since there are 16 independent 4 × 4 matrices. −λ a T a.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 a |a.28) (8. −λ |¯. λ) eik·x + d† (k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ). p.24) (8. λ| a. λ) i γ5 C v(k. λ)v ∗ (k. −p.29) In table 2 the behavior of particle states under the various transformations has been summarized. λ)iγ5 Cu(k. −p.Discrete symmetries 69 Table 8.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k.32) (8.25) (check this for the standard representation). λ)u∗ (k. λ) eik·˜ + d† (k. λ). p.26) d k x x ηT b(k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ)u∗ (k. λ)v ∗ (k. The result is = = λ (8. λ) eik·x + d† (k. λ) Top = ηT b(k. x (8. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. −p. λ) e−ik·x (2π)3 2Ek (8. λ) T −1 = η ∗ d(k.33) (8. λ a P |a.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ). (8. λ |¯. λ| ¯ C |¯.30) (8.31) (8. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. λ). p. ˜ = 0|b(k)|A 8. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). a state |a.34) . λ)iγ5 Cv(k. ˜ Top d(k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. λ)v ∗ (k. p. λ) eik·x + d† (k. λ)u∗ (k. ∗ ˜ v (k.

and T. As an example consider the vector current for a point fermion. ¯ In general the expectation value between momentum states can be more complicated. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P.36) (8.. (8. s . p′ |V µ (x)|p = u(p′ )Γµ (p′ . (8. (q 2 ). Note that the lagrangian density L (x) → L (−x) under Θ. C. γ µ . in this case only q 2 = (p′ − p)2 . s′ |V µ (x)|p. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). called form factors.5 Form factors Currents play an important role in ﬁeld theory.38) (8. C. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ).35) The x-dependence can be accounted for straightforwardly using translation invariance. for the vector current V µ (x). The 16 combinations of Dirac matrices appearing above are linearly independent. p) can be built from any combination of Dirac matrices (1. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. p)u(p) e−i (p−p )·x . This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p .40) where the coeﬃcients are functions of invariants constructed out of the momenta.39) where Γµ (p′ . V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . ¯ ′ ′ (8. as well as under the combined operation Θ = PCT (see Table 3). 2M (8. σ µν . 8. γ ν ].3: The behavior of the independent bilinear combinations of fermi ﬁelds under P.41) 2M 2M .. p′ |V µ (0)|p = u(p′ )γ µ u(p). of which the expectation value between momentum eigenstates can be simply found. 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 . e.g. In many applications the expectation values of currents are needed. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). F. p′ .Discrete symmetries 70 Table 8. F. (q 2 ).37) (8. γ5 γ µ or γ5 ). For instance for nucleons one has Γµ (p′ .

p)γ0 u(p). • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p′ )γ0 u(p). p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p)u(˜) ˜ x p ¯ p p ˜ p (8. p)(i γ5 C).42) where q = p′ − p. p) that Γµ (p′ . p) that γ0 Γµ† (p. p). p)γ0 .1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . µ p ˜ (8. p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. or relations based on hermiticity of the operator or P . or relations that follow from the equations of motion. γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p) that Γµ (p′ . p) = (i γ5 C)Γ∗ (˜′ .g. p) = γ0 Γµ (˜′ . p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . e. the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). e. p)(iγ5 C)u(p). ¯ 2M (8.43) = ei q·x u(p′ )γ0 Γµ (˜′ .45) Exercises Exercise 8. ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p′ )γ0 = Γµ (p′ . Therefore hermiticity implies for the structure of Γµ (p′ .g. p′ )u(p′ ) = u† (p′ )Γµ† (p.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . C and T invariance.Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices.

Exercise 8. 0. 0. |q|). Exercise 8.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element. −|q|/2). (c) Show that hermiticity of the current requires that F1 en F2 are real. p = (E. 0. σµν q ν F3 (q 2 ) 2M . 0. L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . (b) Show that such a term is also not allowed by time-reversal symmetry. GE GM = F1 − Q2 F2 . (b) Show that in the interaction with the electromagnetic ﬁeld. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). 0.Discrete symmetries 72 (a) Use the Dirac equation to prove the Gordon-decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ )[(p′ + p)µ + iσ µν (p′ − p)ν ]u(p) ¯ 2M (b) Give the relation between Hi en Fi . (c) Show that if a term of this form would exist. p′ = (E. 4M 2 = F1 + F2 . |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 .2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. 0.

(9. t0 ) ∂t (9. QH (t)|q. [AH . (9. in which the operators are time-independent. t ≡ q|q. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. in which the states are time-independent. (9.7) 73 . |ψS (t) = U (t.3) (9.6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing. t0 ) = H U (t. AS (t) = AS and the states o are time dependent. t0 )|ψS (t0 ) .2) (ii) Heisenberg picture. t0 ) AH (t0 ) U (t. t0 ). |ψH (t) = |ψH . t′ . i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .Chapter 9 Path integrals and quantum mechanics 9.4) ∂ U (t. ≡ 0. Consider the two (time-independent) Heisenberg states: |q. (9. and the operators are time-dependent.5) (9.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. AH (t) = U −1 (t. t′ ≡ q ′ |q ′ . t ′ QH (t′ )|q ′ . t ′ |q .1 Time evolution as path integral U (t. t . H].

2π (9. which is a hamiltonian function in which the operators are replaced by real-numbered variables.qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . Then qj+1 . QS |q ′ ≡ q ′ |q ′ .14) . with the correction1 being of order (∆τ )2 . tj+1 |qj . tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj . B] + .q .. H(P. [A. The hamiltonian is an operator H = H(P. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . B] + [A. . 2 12 12 (9. ˙ 2π 1 1 1 [A. Q) expressed in terms of the operators P and Q. B]. B]] + [[A. Combining the two terms gives qj+1 . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj .Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . use the Campbell-Baker-Hausdorﬀ formula.8) Choose t as the starting point with |q = |q.16) this. . qj ). .10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . t = q ′ |e−i H(t −t) |q . qj )]) . 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 .13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.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 ′ .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 . tj+1 |qj . t′ |q. (9. 2π (9.qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . .11) (9. n |q (9. |q1 q1 |e−i H∆τ |q . ′ (9. These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|. The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. (9.. t′ |q. eA eB = eC with C = A+B+ (9.12) where the transformation between coordinate and momentum space involves q|p = ei p.

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

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.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions.24) (9.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) .27) (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K.23) Having deﬁned the Gaussian integral.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .25) (see Exercises).21) again a multidimensional integral.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. α(x) = n αn fn and K(x. . 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. A useful property of functional integration is the translation invariance. y) = ∗ m. 2 (9. with Dα F [α] = n dαn N F (αn ) (9. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. (9.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration. Dα F [α] = Dα F [α + β]. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. Consider the gaussian functional as an example. 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 (9. Note that procedure (i) is an example of the more general procedure under (ii). (9.

however. αn = δmn .35) For applications to fermion ﬁelds. (9.29) (9. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.e.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.38) . which vanish.31) δ δ α = dy α(y) = dy δ(x − y) = 1. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). the Gaussian-type functionals. y) θ(y) . αy = δxy .Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . θ2 = 0. δα(x) or in discretized form δ 1 ∂ ∂ with . −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .g.34) (9. δα(x) δα(x) δ αKβ = dy K(x. δα(x) (9. α(y) = δ(x − y).32) (9. In the expansion of the exponential (from second to third line). (9. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. i. n θn fn (x) dx θ∗ (x) θ(x) . θη = −ηθ. e. (9.36) 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. dx dy θ∗ (x) K(x.30) (9.33) (9. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . products of the same pair appear. F [θ. In principle the deﬁnitions of functionals is the same. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. Examples of functional diﬀerentiation are (9.

Path integrals and quantum mechanics

78

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

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

= δ(x − y),

(9.39)

(9.40) (9.41)

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

9.3

Time ordered products of operators and path integrals

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

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

q ′ , t′ |q, t =

Dq exp i

t

dτ L(q, q) , ˙

(9.44)

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

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

t′

Dq exp i

t i 2

dτ Leﬀ (q, q) , ˙

(9.45)

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

i

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

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

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

t′ t′ t

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i

t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

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

t t′

(9.49)

dτ L(q, q) . ˙

(9.50)

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

9.4

An application: time-dependent perturbation theory

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

′

(9.51) (9.52) (9.53)

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

t′ ∞ n=0 t′

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),

t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

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

**Path integrals and quantum mechanics
**

t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

′

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

t′

U

(n)

(t , t) = (−i)

t

′

dτ HI (τ ) U (n−1) (τ, t)

t′ τ1 τn−1

= (−i)n

t

dτ1

t t′ t′

dτ2 . . .

t t′

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

t

=

(−i)n n!

dτ1

t t

dτ2 . . .

(9.60)

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

t′ t′

+

1 2

dτ2

t τ2

dτ1 HI (τ1 ) HI (τ2 ),

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

t′ t′

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )

t′

+

1 2

t′

dτ2

t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

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

t′ t′

dτ1

t t

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

the desired result.

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

∞ t′ t′ t′

dτ1

t t′ t

dτ2 . . .

t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .

t

say.s. t) φ∗ (Q) e+i En T . i. L(q. The Heisenberg state |q. the creation of an electron (think of a radio-tube. ˙ ˙ physically pictured as. t′ |q. q) + J(τ )q] . t|n = q|n e−iEn t with q|n the time-independent wave function. making the electron) and the absorption of an electron (think of a detector). t′ |q. t|n n|Q. T < t < t′ < T ′ . Before and after these processes there is only the vacuum or ground-state |0 . t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. t is related to the Schr¨dinger state |q by |q. q) → L(q. ˙ J (9. for g.62) p Dq D exp i 2π t dτ [pq − H(p. T ′ |q ′ . q) + J(t) · q.65) J dq ′ dq Q′ . q)] ˙ (9. t q.66) . T = n q. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. t|Q. T ′ |Q.64) (9.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result.e. q .Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. t = = ′ ′ V = q . t |q. it is possible to switch on an interaction. harmonic oscillator. t Q′ . Considering the source to be present between times t and t′ . t|Q. φn (q. i. a set |n of physical eigenstates. T with q.e. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. t′ q ′ . one has q ′ . furthermore.5 The generating functional for time ordered products By introducing a source-term. which in turn are embedded between an early time T and a future time T ′ . Consider. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. T .63) for plane waves. n (9. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). q)] . for the physical states o q. 9. t = ei Ht |q . T = n J t′ = = Dq exp i t dτ [L(q.

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

when L q. The diﬀerentiations in Z[J] and ZE [J] are related. (9. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. det K Exercise 9. complex. δ n Z[J] 1 Z[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. ZE [J] = = where LE q. .Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional.73) d˜ LE q. δJ(tn ) = (i)n J=0 1 δ n ZE [J] .75) = − LE q. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q).72) (9. . i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ .2 Check the examples of functional derivation for real-. ˜ ˜ ZE [J] δJ(t1 ) . Exercises Exercise 9. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. ˜ dt (9. for Grassmann valued functions.and Grassmann-valued functions. 1 = √ . δJ(tn ) J = 0 ˜ t = it (9. y) = ∗ m. det K 1 .74) − V (q) 2 (9. where Kp are the eigenvalues of the matrix Kmn in K(x. . ensuring convergence for the functional integral ZE [J].1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. τ dq ˜ dt ≡ −L q.g. i dq . + V (q). (9. e. .n fm (x) Kmn fn (y). .76) which is a positive deﬁnite quantity. 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 .

Path integrals and quantum mechanics 84 Exercise 9.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω .79) (9. y) K −1 (y. det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η . 2 (9.78) (9. . where dy K(x. z) = δ(x − z).3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω .

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

which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . . As discussed in the previous section (exercise 9. xn ) = x2 xn . . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10.6) the generating functional is (using ∂µ = − ∂µ ) given by Z0 [J] = Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). J(xn ). one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. (10. .11) Depending on the path choosen in the complex k 0 plane (see exercise 7.e. = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10.3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . (10. . . (10.Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). . with Z0 [0] = 1. d4 xn G(n) (x1 . . .9) i.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. 2 2 → (10.13) . Furthermore. . xn ) J(x1 ) .5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . . These Green functions appear in the expansion Z[J] = n in n! d4 x1 .10) J=0 In order to determine ∆F . . (2π)4 (k 0 + iǫ)2 − E 2 (10.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x .

15) e−i k·x d4 k . i. 4 k 2 − M 2 + iǫ (2π) (10.18) Thus we see various solutions. the advanced Green’s function ∆A (x) = satisfying ∆A (x. 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 . (10.17) d4 k e−i k·x . t) = 0 for t > 0. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . µ 1 1 • Firstly.e. (2π)4 k 2 − M 2 (10. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x).1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).16) C where C is a closed contour in the complex k 0 -plane.20) = exp − 2 where or (see ﬁg. φ+ (0)] = (2π)3 2E are also shown in exercise (7.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x .22) (10. . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). ∆(x) = − d4 k e−i k·x . (2π)4 k 2 − M 2 (10. t) = 0 for t < 0.14) They are solutions of the inhomogeneous equation.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . ∆F = ∆C . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. 10. (2π)4 (k 0 − iǫ)2 − E 2 (10. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . 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).4). In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. 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).

x3 . . we have G0 (x1 . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) .25) x1 x2 x3 = x4 + + . = θ(x0 − y 0 ) 0|[φ+ (x). 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.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). δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . or diagrammatically (4) (10. because this is the only quantity entering Z0 [J].Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. . the same result as above. We ﬁnd (see also section 7.1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility.24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). x2 .

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 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.30) and in zeroth order G(0.32) (10.g 0 ) = G0 = 1. (2) (0) (10. x2 .31) G(2. x2 ) = i ∆F (x1 − x2 ) = G(4.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. x2 . x2 ) = G0 (x1 . with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J]. x4 ) = G0 (x1 .g ) (x1 . the generating functional can be written. (10. (10. x3 .35) z . L (φ) = L0 (φ) + LI (φ).29) This expression will be the one from which Feynman rules will be derived.g ) (x1 . i. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . Z[J] = = Dφ exp i exp i d4 z LI (10.Feynman diagrams for scattering amplitudes 89 10.e. x3 .27) When interactions are present.2 Generating functionals for interacting scalar ﬁelds d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10. (10.34) Introducing a vertex point to which four propagators are connected.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 . ≡ −i g d4 z (10.

. . . J(xn ) c (10. x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . G(n) (x1 . . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) .39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. . .e. J=0 (10. (10. xn ) J(x1 ) . . . (10. d4 xn G(n) (x1 . (10. . . Diagrammatically this exponent only contained the two-point function. which also was the only Green’s function for which the diagram was connected.g 1 ) = 1 8 z 1 8 . Z[J] = n in n! d4 x1 . . δJ(xn ) . . .37) G(2. (10. . This remains also true for the interacting case.38) The result for the 4-points Green’s function is left as an exercise. d4 xn G(n) (x1 . To see this write Z[J] = exp (i W [J]) with (by deﬁnition) the expansion W [J] = n or i W [J] = ln Z[J]. . i. . did not contain parts that could be written as products of simpler Green’s functions. xn ) J(x1 ) .36) J J from which one obtains G(0.41) in terms of socalled connected Green’s functions. x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . .g ) (x1 . .40) in−1 n! d4 x1 . . . In the free case we have seen that the exponent of Z[J] contained all essential information. J(xn ). . . .42) x1 G(n) (x1 . Connected Green’s functions We have now seen the Green’s functions as the quantities appearing in the expansion of the generating functional Z[J]. xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . . . (10.Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J].

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 + + ... (iv) If δZ[J] = i 0|φ(x)|0 = 0. in the expansion of which all vacuum blobs are contained. (1) (1) (iii) Gsc = Gc .. 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. + +i d4 x1 1 2! + 1 3! +i d4 x1 + 1 2! i2 2! d4 x1 d4 x2 + . 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. 1 2! Z[0] = 1 + + + . In the expansion of Z[J]/Z[0] one (by deﬁnition) has the socalled source (n) connected Green’s functions Gsc ..Feynman diagrams for scattering amplitudes 91 To see that Z[J] = exp(i W [J]) contains all non-connected diagrams can be performed inductively....

this will be translated to the action on ﬁelds. in = |p1 . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). then (n) (n) Gsc = Gc for n ≤ 3.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. translation invariance). Next. (2) (10. . (10. out = δαβ ↔ SS † = 1. pn . in = β. out| and S † |α.47) . in|p′ . e. in = |α.e. in|S = α. For the free scalar theory. in = p. out| ⇔ |α. . x2 . t = −∞ φin (x) t = +∞ φout (x) φ(x) where φin and φout transform under the Poincar´ group as scalar ﬁelds and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . in| (choose ϕ0 = 0).Feynman diagrams for scattering amplitudes 92 i. out (suppressing except momenta all other 1 m quantum numbers). in into ﬁnal state particle states (out-states) |β. considered explicitly. in = p. . ∂µ ∂ µ + M 2 φin (x) = 0. x1 G(4. in = S|α.g ) (x1 . one ﬁnds for the connected 4-point Green’s function at order g. x4 ) = c 1 x3 z (10. thus β. Proof: α. p′ . (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). . (2) The one-particle state is invariant (conservation of energy and momentum.g. (10. φout and the interpolating ﬁeld φ(x). out = |p′ . out . p. out|p′ . . x4 10. out|α. out|p′ . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . out = p|p′ . . in|SS † |α. Sβα ≡ β. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . implying the absence of tadpoles. in|S = 0.45) x2 as the only surviving diagram (see exercises). Proof: 0. in|S|p′ . in|S = β. out|α.44) J=0 For the interacting theory. in ⇔ β. (10.43) 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 ).3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. x3 . in = p. out . out| = eiϕ0 0. the vacuum expectation value of the ﬁeld φ(x) is zero.

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 ).50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . z) φ(z) d4 y d4 z ∆A (x − y) K(y. The convergence therefore must be weakened to t→−∞ √ (10.51) (10. in|φin S → φin S = S φout (x).54) . out|φout = β.and out-ﬁelds is: φin (x) = S φout (x)S −1 . this can actually not be used as it would imply [φ(x). (2π)3 2E (10. the source term Jφ in the lagrangian density is treated in the interaction picture. a causality condition that can be proven to imply the absence of interactions. (10. implies the strong (operator) convergence φ(x) → Zφin (x). e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . The relation between S-matrix and in. φ(y)] = Z [φin (x). e. a) = S. √ Although the above. The time evolution operator (see Eq. which in a later stage (renormalization) will become more important.52) where Z is a constant.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). Proof: β. 9. (10.g. a)SU −1 (Λ. z) φ(z). 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. 2 ∂µ ∂ µ + M0 φ(x) = J(x). The relation between S and Z[J] To establish this relation.e. β.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. i.48) The fact that φin and φout satisfy the homogeneous Klein-Gordon equation with the physical mass M implies that they create particles and antiparticles as discussed. φin (y)] = iZ ∆(x − y). (10. as it stands. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. S is invariant under e Poincar´ transformations: U (Λ. 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.

φ(x). [A. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. φ(y)] f (y) e e . express it in terms of advanced and retarded Green’s functions. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. S U [J]] = √ Z δ S U [J].58) In order to ﬁnd a solution to this equation note that. e = i. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. C] are c-numbers. δJ(z) (10. B] is a c-number).59) (10.e. just as we did for φ(x). t i Ut∞ [J] φ(x)U−∞ [J] (10. δJ(z) (10.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x).57) i [φin (x).63) . z) d4 y d4 z ∆(x − y)K(y. e e ] = [A. one has [A. B] (for the case that [A. y) δ δJ(y) : F [J].61) e φ+ f where the normal ordered expression : e φ f : is used. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). z) δU [J] . B + C]e e . (10. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. φ(x). which is equal to the expression e in which the creation operators are placed left of annihilation operators. B]eB . with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. provided [A. (10. δJ(z) Taking the diﬀerence between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. eB ] = [A. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J].55) δU [J] δJ(x) = t→∞ t→−∞ −→ −→ √ i φ(x) U [J] → i Z φout (x) U [J] √ i U [J] φ(x) → i Z U [J] φin (x). 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. z) . φ− f (10. we can. B] and [A.62) where F [J] is some (arbitrary) functional.

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

If problems arise go back to the deﬁning expression for the generating function in 10.27. or calculating full Green’s functions external lines are treated as propagators. which cancels the factor 1/2 in the quadratic piece. but overall momentum conservation at a vertex is understood. ’t Hooft and M. 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. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically diﬀerent connected diagrams.27. Veltman. (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). = −i g . At these vertices each line can be assigned a momentum. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles).66 and 10. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. the factor 2 corresponds to the two-points Green’s function having two identical ends). 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. For the symmetry factor consider the examples (given in G. Diagrammar) in the case of the interaction terms g f (10. vertices and external particle wave functions in that diagram. Note that in calculating amputated Green’s functions external lines are neglected.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. For the interaction terms in the lagrangian.69) LI (φ) = − φ3 − φ4 .

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

Feynman diagrams for scattering amplitudes 98 Dirac ﬁelds For fermions the generating functional is given by Z[η. e.73) = i p / − M + iǫ = ji i(/ + M )ji p p2 − M 2 + iǫ The time ordered functions are obtained by functional derivatives from Z(η.e. 4 2 − M 2 + iǫ (2π) p and the (oriented) propagator in Feynman diagrams involving fermions is i p j (10. (10. 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 . − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ )... η]. The latter rule is illustrated in the example of an interaction term LI = : gψ(x)ψ(x)φ(x) : in an interacting theory with fermionic and scalar ﬁelds. . = + .g. in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is (see next section for e-e-γ vertex). which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. which arises from the quadratic term in exp i d4 z LI .− i δη(z) i δη(z) d4 x d4 y η(x) iSF (x − y) η(y) .72) (10. The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution. η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ .71) = exp i where iSF is the Feynman propagator for fermions.. but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams. η). − g2 2 dz dz ′ δ2 δ2 δη(z) δη(z) δη(z ′ ) δη(z ′ ) and the quadratic term in Z0 [η. (Rule 6) Each closed fermion loop gets a sign −1. namely (Rule 5) Feynman diagrams which only diﬀer by exchanging identical fermions in initial or ﬁnal state have a relative minus sign.

As an example some scattering processes in quantum electrodynamics will be discussed in the next section after the introduction of the Feynman rules for vector ﬁelds. λ 1 ∂ (10. 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 . 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. writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. The wave functions are given by µ k ǫµ (k) k µ incoming photon outgoing photon ǫ∗ (k) µ The vertex for the coupling of photon to the electron is given by µ = i j −i e(γµ )ji .Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. + iǫ)2 where the kµ kν terms in the case where the photon couples to a conserved current (such as ψγ µ ψ) will not contribute. Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). Vector ﬁelds and Quantum Electrodynamics As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). λ (∂µ Aµ )2 2 1 For a massive vector boson inversion of the quadratic term including the Lorentz constraint ∝ the propagator kµ kν 1 −gµν + 1− i Dµν (k) = i . i p p i ui (p) ui (p) ¯ vi (p) ¯ incoming fermion outgoing fermion incoming antifermion outgoing antifermion i p p i vi (p) In writing down the expressions for Feynman diagrams one has to be aware that the wave functions and the propagators have more than one component.74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ .

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

s1 )γν v(k ′ . the masses. s4 )γ ν u(p. Exercises Exercise 10. s2 (10.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. s2 ) ¯ ¯ 1 2s u(p. s3 )γ µ v(p′ . (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. (c) Do the same for the connected Green function Gc . To lowest order (∝ α = e2 /4π) only one diagram contributes.2. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. This is known as crossing symmetry. s2 )γµ u(k. x4 ) for the interacting case in φ4 -theory to ﬁrst order in the coupling constant g as obtained from the full expression for Z[J] in section 10.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.s3 . (4) (4) . Similarly.s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . The diagram. .1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . s2 )γµ u(k.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ .s2 3 . s1 )|2 ¯ s 1 . s4 ) u e2 v (k ′ . s3 )γ µ v(p′ .s2 . s1 ) u(k. . s3 ) ¯ ¯ 1 . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. . s4 ) v (p′ .

. .38. . . their sum is zero. + pn ). Show that i the amplitude is symmetric under the interchange of t ↔ u. Pictorially this implies. . .4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. It is of the form −i M = A1 − A2 . or if we in general would consider all momenta as incoming.3 Show that the combinatorial factors found using the rules given in section 10. . . which means overall momentum conservation in Green functions in momentum space. |M |2 = T11 + T22 − T12 − T21 . . xn ) ≡ (2π)4 δ 4 (p1 + . . + pn ) G(n) (p1 . . . . . t and u. 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). Exercise 10. . hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. . xn ) ∝ (2π)4 δ 4 (p1 + . . . 1 2 (b) Calculate the quadratic pieces and interference terms. . . . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . . Express the contributions in invariants s.37 and 10. Exercise 10. . . pn ) only has n − 1 independent variables in it. xn + a) = G(n) (x1 . . pn ). Note that the function G(n) (p1 . in the amplitude (Tij = A∗ Aj ). but the incoming and outgoing momenta are identical. xn ).Feynman diagrams for scattering amplitudes 102 Exercise 10.2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. . for instance. e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . .

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

m2 . which in the speciﬁc case a b also can be expressed as λ(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. (11. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . m2 ) = 4(pa · pb )2 − 4p2 p2 .10) The function λ(s.16) . p1 . m2 ).8) (11.15) Explicit calculation of the reduced two-body phase space element gives dR(s.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . = 2 s λ(s. p2 ) = CM = = 1 (2π)2 1 (2π)2 1 (2π)2 d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2 which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b .9) (11. p2 ) = = where λ12 denotes λ(s. m2 . m2 ) is a function symmetric in its three arguments. m2 . p1 . 2mb λ(s. In that case trf pa = (Ea . 4s (11. 2mb (11.12) pb = (mb . m2 . 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ q |q| dΩ(ˆ) λ12 dΩ(ˆ) q √ = .11) (11. 0). m2 ) a b . mb and s). for instance. m2 .13) (11.14) One can. a (11. ptrf ). Also in this case one can express the energy and momentum in the invariants. 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 ) a b . Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . 2 s s − m2 + m2 a b √ . 4π s 4π 8π s 4π (11.

11. pc .4). tmax min λcd /4s. pd ) = 4π s 4π 8π s 4π dt dt √ √ . The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . 2s (11. The various possibilities are referred to as diﬀerent reaction channels. An often used set of invariants are the Mandelstam variables. At high energies. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt.25) dR(s.17) (11. (11. in exclusive measurements all particles are detected.. Consider the process a + b → c + d. The initial state.22) (11. 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 . Consider the 1-particle inclusive case. there is usually a preferred direction. in which one particle is detected.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .19) (11.20) → π0 + n → π+ + π− + n → . usually is a 2-particle state. pz = mT sinh y. a c with q = λab /4s and q ′ = being 0 or 180 degrees. H1 + H2 → h + X. Examples appear in → π− + p (11. however.21) (11. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. E . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . usually many particles are produced.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. pT = (px . while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). In inclusive measurements no particles are detected in the ﬁnal state.. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. = (11. . Instead of the scattering angle. The variable s is always larger than a c b d the minimal value (ma + mb )2 .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes.27) with m2 = m2 − p2 = m2 + p2 + p2 . Of course there are not only 2-particle channels.18) (11. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. py ). Writing E = mT cosh y. A speciﬁc reaction channel starts contributing at the threshold value (Eq.

uu = (pa − p¯)2 = s.2 Crossing symmetry In the previous chapter. u). This is known as crossing symmetry. uu = s) = Mab→cd (s. 11.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β.31) . 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 = = = .30) (11. t-channel and u-channel processes respectively. ¯ b (11.28) which (only involving angles) is easier to determine. (11. t. c bd Mad→c¯(su = u. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). b ut = (pa − pd )2 = u. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. u). ¯ tu = (pa − pc )2 = t.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. (11. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. ut = u) = Mab→cd (s. ¯ tt = (pa − p¯)2 = s. u = (p1 − p4 )2 = (p2 − p3 )2 . tt = s. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. tu = s. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. η = − ln(tan(θ/2)) = tanh−1 (cos θ). we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. t. 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. t. which means that rapidity distributions dN/dy maintain their shape. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . t = (p1 − p3 )2 = (p2 − p4 )2 .

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

From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. dt (11. (11. T · V ρa (11.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). 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 . as expected.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. .e.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. 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. (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.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ).34) i.38) (in which we can calculate −iMf i using Feynman diagrams). pn ).36) (11. it is. Using ρtrf = a (0) (0) (0) lab lab lab ρa · γa = ρa · Ea /ma (where ρa is the rest frame density) and va = plab /Ea we have a trf ρa ρb va = (0) ρa ρb 2 λab 4 ma mb (0) (0) or with ρa = 2 ma . the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . N and T · V are covariant.35) This quantity is not covariant. . the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. N 1 . For the (proper) decay width one thus has 1 N Γ0 = . (11. . σ= √ 2 λab T · V (11.T V. .Scattering theory 108 Although this at ﬁrst sight does not look covariant. .

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

the total cross section can be determined. 0). (11. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . (11. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 .49) or Im Mℓ = q |Mℓ |2 .48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .52) From the imaginary part of M (s.53) The diﬀerence is the inelastic cross section.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 .50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. (11. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni . parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). Using σ= dΩ q ′ M (s. θcm ) √ q 8π s 2 .54) ℓ . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . (11. ℓ (11. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). 2i q 2i q (11. In general a given channel has |Sℓ (s)| ≤ 1. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ).e.51) and for the case that this is the only channel (purely elastic scattering. n (11. q ˆ i.

To check unitarity we need the amplitude at order e4 .5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. This is what we will do to discuss decay widths. the time-evolution operator.56) (again disregarding spin).58) . . 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 . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. 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. 0) ∝ e−i(E−iΓ/2)t . We note that unitarity is generally broken in a ﬁnite order calculation. (q Mℓ )ij (s) = −M Γi Γj . The prescription for the pole structure. 0)|2 = e−Γt . s − M 2 + iM Γ (11. which is given by a sum of the partial widths into the diﬀerent channels. The product of amplitudes is of order e4 . (11. 11. . For instance the tree-level calculation of Feynman diagrams at order e2 was real. however. 0). pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. speciﬁcally one has for t > 0 that the Fourier transform is dk e 0 −ik0 t ∆(k) ∝ e−ik t dk (k 0 − Ek + iǫ)(k 0 + Ek − iǫ) 0 0 (t>0) ∝ e−iEk t . For the amplitude in a scattering process going through a resonance. i. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . i. such that |U (t. ∝ U (t. We can also invert the situation and use unitarity to ﬁnd the imaginary part of an amplitude at some order from a lower order calculation.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption.e. in which case. σel = σinel .55) (note that we have disregarded spin). For an unstable particle one expects that U (t. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. η = 0.e. but still real. it is straightforward to write down the partial wave amplitude.57) This shows that Γ is the width of the resonance.

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

000 123 Γ. Exercise 11. writing N (s) M1 (s) = 2 + iM Γ .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0.Scattering theory 113 Exercise 11. ρcm ρcm a b 1 cm .3 Complete the calculation of the decay width and lifetime of the pion by including the appropriate phase space factors and using the calculation of the squared matrix element in Exc. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . Thus. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. cm |va − vb | in the center of mass system for two colliding particles a and b. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width.999 877 Γ and Γ(π − → e− νe ) = 0. 4. a quantity which we will ¯ ¯ encounter in the next chapter (section 12.8. Calculate now fπ . Compare the results with the experimental π − lifetime τ = 2.2 Calculate the ﬂux factor. 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 11.2). As discussed for unstable particles.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∆ .

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

What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop. In principle such a phase in a given point is not observable. but expresses them with respect to diﬀerent bases. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). = . (12. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). However. -dx -dy µ µ dyµ µ dx For a vector. 12. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. should be independent of such transformations.18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. 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). Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). 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 .The standard model the basis for x G and x+dx 117 G. i.16. (12. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. (12.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). (12. ds which is solved by dψ(s) ds = dxµ . i. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x).19) This gives rise to a (path dependent) phase in each point. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).e.e.14) (12.15) ψ(x). 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. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). which connects two identical vectors. The relation in Eq.

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

(12. etc. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . while the lagrangian including the nonzero groundstate expectation value chosen by nature.e. The result is 1 1 1 1 (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 − m2 (ϕ2 + ϕ2 ) L = 1 2 2 2 2 2 1 2 1 − m (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12. while the quadratic terms must correspond with non-negative masses.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). . Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. has less symmetry. ϕ2 and η. for the vacuum Qa |0 = 0. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. In this situation one speaks of spontaneous symmetry breaking. It is only invariant under rotations around the 3-axis.22) F Realization of symmetries (model independent) In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented. (12.25) 2 2 2 Therefore there are 2 massless scalar particles. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . ϕc = 0|φ |0 = (12. 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 . φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0.The standard model 119 a constant ﬁeld (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2. In this case the spectrum is described . It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic.) terms constitute interaction terms. They are known as the Weyl mode or the Goldstone mode: Weyl mode. Any constant ﬁeld with ’length’ F is a possible groundstate. .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. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. (12. The classical groundstate appears degenerate. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . therefore a choice must be made. The situation now is the following. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . . In the case of degeneracy. i. The η massless particles are called Goldstone bosons.23) F +η the latter only forming a minimum for m2 < 0. say 0 0 .

i. This means that they are operators that create states from the vacuum. 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. ∂ ∂ (12. ψ −→ ei α·T ψ. etc. denoted |π a (k) . [Qa .27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. a massless Goldstone boson for each ’broken’ generator. irrespective of the fact if they annihilate the vacuum. a and a′ are degenerate states. . Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x . As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. Goldstone mode. H] = 0. a ∂ µ Jµ = 0.26) for all the states labeled by a corresponding to ’broken’ generators.e.. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. strange.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). Neglecting at this point the local color symmetry. called ﬂavors. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). Taking the derivative.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab (fπ = 0) (12. . ψu mu md ψ = ψd . In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. if the generators Qa generate a symmetry. . (12.The standard model 120 by degenerate representations of the symmetry group. Using the Dirac equation. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. it is easy to see that one gets ∂µ V µ = i ψ [M.) one can write L = f ψ f (i/ − Mf )ψf . . and one must have mπa = 0. i. (12. V µ = ψγ µ T ψ.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor).32) (12. T = τ /2. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . T ] ψ. there must be a nonzero expectation value 0|Jµ (x)|π (k) .e. π (12. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .e. The generators Qa (in that case the rotation operators Lz . (12. i. down. A bit more formal. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. Including a sum over the diﬀerent ﬂavors (up. M = .31) . = ψ(i/ − M )ψ.

There exists another set of symmetry transformations. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. ψR/L = 1 (1 ± γ5 )ψ.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. In the real world. it is easy to see that one gets ∂µ Aµ = i ψ {M. generated by TR/L = 1 (1 ± γ5 )T . e.27.3 MeV/c2 and Mn = 939. but also a triplet (isospin 1) of pions. ψ −→ ei α·T γ5 ψ.e. 12. i. I. i. but the fact that the symmetry is not perfect and the right hand side of Eq. 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.35 is nonzero. while the groundstate is only invariant under isospin rotations (R = L). Still the basic fact that the generators acting on the vacuum give a nonzero result (i. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . which is equivalent 2 to the V-A symmetry. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry.34) Using the Dirac equation. From the number of broken generators it is clear that one expects three massless Goldstone bosons. parity minus). This combined symmetry is what one calls chiral symmetry. parity would not play a role in the world. (leaving out the ﬂavor structure) the same as ψγ5 ψ. much smaller than any of the other mesons or baryons. proton and neutron. which is approximately true for the up and down quarks. stated equivalently. the lagrangian density is invariant under left (L) and right (R) rotations independently. T } γ5 ψ. ∂ ∂ (12. T ] = 0. Therefore the world of up and down quarks describing pions. (12. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . is by nature not realized in the Weyl mode. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. fπ = 0 remains. socalled axial vector transformations. if in ﬂavor space M is proportional to the unit matrix. This situation. both are very small. including in Dirac space a γ5 matrix.36) If the mass is zero the lagrangian is split into two disjunct parts for L and R showing that there is a direct product SU (2)L ⊗ SU (2)R symmetry. chiral symmetry is not perfect. (12. The chiral ﬁelds ψR and ψL are transformed into each other under parity.e.e. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . etc.e. as the quantity ∂µ Aµ (x).35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. or. behaves as a pseudoscalar particle (spin zero. Aµ = ψγ µ T γ5 ψ.g. with almost degenerate masses (Mp = 938. however. is what happens in the real world. T γ5 = τ γ5 /2. (12. M = m · 1. How can we see this. for which the ﬁeld (according to the discussion above) has the same behavior under parity. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. 12. From group theory (Schur’s theorem) one knows that the latter can only be true. Note that these transformations also work on the spinor indices. i.e. This symmetry.6 MeV/c2 ).33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. where the quark masses are not completely zero. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . . etc. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. a doublet (isospin 1/2) of nucleons. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical.

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

the quarks (up. while the lefthanded particles form a doublet. electromagnetic and weak forces. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. down. the leptons (elektrons.) and the gauge bosons responsible for the strong. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. (12. however. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . ψR/L = 2 (1 ± γ5 )ψ decouple. This is written as 3 Q = TW + YW .51) . ∂ ∂ ∂ (12. νµ and ντ ) and their antiparticles. τ ). left.45) (12.e. = ∂µ L − (12.The standard model 123 12. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. etc.50) (12. muons. 2 (12.48) (12. that gives a good description of the physical world.). neutrinos. explicitly L −→ e−i β/2 L. (12. 1 As they are massless. 2 2 = ∂µ R + i g ′ Bµ R.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. R.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. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained.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. muon µ− or tau τ − ). under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . R −→ e−i β R.47) and YW is considered as an operator that generates a U (1)Y symmetry. i. its corresponding neutrino (νe . In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. The procedure. U(1)Y U(1)Y (12.and righthanded particles. 0 1 +1/2 L νe 3 − = with TW = and TW = L= −1/2 eL L 2 R = R − = eR 3 with TW = 0 and TW = 0.44) which has an SU (2)W symmetry under transformations eiα·TW . Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. etc.46) −→ 3 One notes that the charges of the leptons can be obtained as Q = TW − 1/2 for lefthanded particles 3 and Q = TW − 1 for righthanded particles. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.

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. leading to the parametrization 1 0 φ(x) = √ (12. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . (12. the U (1)Q symmetry (generated by the charge operator).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. 4 4 In order to break the symmetry to the symmetry of the physical world. † i i g Wµ · τ − g ′ Bµ )φ. L where L (h1) (h) =L (h1) +L (h2) .59) .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. . Using local gauge invariance θi for i = 1. For the classical ﬁeld the choice θ4 = v is made. .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .52) L L (f 1) (f 2) = = i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ + i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 . .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. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. (12. (12. Putting this in leads to L (f ) =L (f 1) +L (f 2) .53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. h) (12.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1. this leads to the lagrangian L (h1) . . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. −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 ). which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. .58) (12.

4 2 MW g2 v2 = .64) (12.61) (12.65) The weak mixing angle is related to the ratio of coupling constants. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 2θ 2θ 4 cos W cos W 0. (12. 2 Zµ (12.63) (12. 2 (12. . (12. g ′ /g = tan θW .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 . we can read oﬀ e = g sin θW = g ′ cos θW .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ . for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + . 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 . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12. .The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ .67) The coupling of electrons or muons to their respective neutrinos. 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 .66) From the coupling to the photon.69) . σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.60) (12.

231 2. (12.166 4 × 10 0. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. i.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. g ′ and v determine a number of experimentally measurable quantities.e.e. such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. Finally we want to say something about the weak properties of the quarks. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). 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. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. 1.g. (12.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. Comparing this with the total width into (invisible!) channels. 91. −5 (12.72) (12. me /v is extremely small. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. . For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.19 GeV. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).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 . giving rise to a running coupling constant after renormalization of the ﬁeld theory.71) GeV −2 = = = = .78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. g2 MZ 2 2 (CV + CA ). one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. muon or tau. v (12.77) (12. i. 3 = TW .73) (12. from the MW mass is about 250 GeV. 2 2 2 (12.76) with CV CA 3 = TW − 2 sin2 θW Q. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).40 GeV. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.5 MeV (exp.80) First.74) (12. The coupling to the Higgs particle is weak as the value for v calculated e. one has the relation g2 e2 1 GF √ = . 80. With the choosen vacuum expectation value for the Higgs ﬁeld. (12. (12.

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

respectively).86) the unitary CKM-matrix introduced above in an ad hoc way.82 and ρ ≈ 0. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . each of these containing the three families. ms and mb ). d d (12. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. (12. 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 . u u where we include now the three lefthanded quark doublets in QL .82) space.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . (12. . Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. e. UR = uR cR tR .34.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2. † Λu Λ† = Vu G2 Vu .227. A ≈ 0.22 and η ≈ 0.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . 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. 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 . for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .g.83) where Vu and Vd are unitary matrices. i. the weak mass eigenstates are L (h2. The Higgs ﬁeld is still limited to one complex doublet.e.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .

it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). one can write Λe = Ve Ge We and we ﬁnd † L (h2. † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . The mass ﬁelds ER † mass † mass = We ER . 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 . NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass .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 . since the weak current is the only place where they show up.5×10−3 eV2 . and obtain L (h2. EL = Ve EL . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. 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.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2.92) .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me . EL = Ve† EL . decoupling from all known interactions.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. In this case.e. mµ and mτ . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. (12.89) with three righthanded neutrinos added to the previous case. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . (12.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .ℓ) = −LφΛe ER − ER Λ† φ† L. (12.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. e N L= L EL (12. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL .90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . there is no family mixing for massless neutrinos. we ﬁnd massless neutrinos. As before.93) 1 = 0 0 0 c23 −s23 0 s23 c23 c13 0 −s13 eiδ 0 s13 eiδ 1 0 0 c13 c12 −s 12 0 s12 c12 0 0 0 1 . (12. i. UPMNS a parametrization that in principle also could have been used for quarks.

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

(12.1 Consider the case of the Weyl mode for symmetries.The standard model 131 taking ML and MR real and non-negative.2 Derive for the vector and axial vector currents. i.e.105) for i = 1.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . i.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. 2 U (M M † ) U † = M0 . a and a′ are degenerate states. related via c Υ1R = U nL .101) Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2 .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)) . [Qa . (12. L Υ2L nR Υ1 and Υ2 . 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′ . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . i ψ {M.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 . Exercise 12. T ] ψ. T } γ5 ψ. H] = 0. Exercise 12. (12. 2 with real masses Mi . 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). Prove that if the generators Qa generate a symmetry. Υ2R nR (12. Exercises Exercise 12. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. .e. 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 . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates.

ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). etc.231.5 Show that the couplings to the Higgs (for W + W − h. Exercise 12. e+ e− h. the weak mixing angle is given by ¯ sin2 θW = 0. Exercise 12. mb ≈ 5 GeV and MW ≈ 80 GeV. The mass of the Z 0 is MZ = 91 GeV. Exercise 12.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . Mf M2 and ghf f = . ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld.7 In this exercise two limits are investigated for the two-Majorana case. 3 TW . (b) Calculate the width to electron-positron pair. to be precise one has for bosons (b) or fermions (f). and the width to a pair of neutrino’s. Γ(Z 0 → νe νe ). Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 . Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. Γ(Z 0 → e+ e− ). ZZh.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . hhh. at least up to a high (which?) order in λ. . Exercise 12.) are proportional to the mass squared for bosons or mass for fermions of the particles.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . 2 2 The masses are mt ≈ 175 GeV. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 .

which leads to the socalled see-saw mechanism. . Calculate the eigenvalues ML = 0 and MR = MX . 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. (b) A more interesting situation is 0 = ML < |MD | ≪ MR . Given that neutrino masses are of the order of 0.05 eV. what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV). Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 .The standard model 133 (a) Calculate for the special choice ML = MR = 0 and MD real.

relativistic quantum mechanics

relativistic quantum mechanics

- quantum field theory
- Quantum Field Theory 2013
- Mandl F., Shaw G. - Quantum Field Theory
- Huang K. - Quantum Field Theory
- Introduction to quantum field theory, based on the book by Peskin and Schroeder
- PESKIN - Quantum Field Theory
- Srednicki M. Quantum Field Theory - Solutions Manual(Web Draft, 2006)(156s)_PQtb
- An Introduction to Quantum Field Theory - Peskin
- An Introduction to Quantum Field Theory(Peskin)
- Srednicki Ms Qft Solutions Rev
- Weinberg, Steven - The Quantum Theory of Fields Volume II
- Lahiri and Pal-A-First Book of Quantum Field Theory-2nd Ed
- Weinberg the Quantum Theory of Fields Vol 1 Foundations
- Phy 331 Advanced elsctrodynamics.pdf
- Conductor2.pdf
- Advanced
- lec38
- Advanced quantum mechanics.pdf
- set9
- Advanced Quantum Mechanics
- Advanced Quantum Mehanics-Peter S. Rise Borough
- part4 kjkj
- 1)Physical Quantities and Units (1)
- Electromagnetic Field Theory Exercises- Bo Thide, Upsilon Books 2001
- Additional Mathematics Project Work
- QuaternionsI
- 00019394
- Avec Wave
- Dpp (6) 11th Physics_Eng_WA Hw
- Submarine Control
- QFT-0

Are you sure?

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

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

Get the full title to continue

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

scribd