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

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

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

November 2011 (version 6.04)

Contents

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

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

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

1

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

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

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

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

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

Vol.1 1.2 1.3 3.7 2.-B. 1980.3 2.1. These notes Section 1. 2.2 3.2 3.3 6.3 2. 1996.E. Peskin & Schroeder and Srednicki. Cambridge University Press. Peskin and D.4 6.5 2. 1995.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. M. Addison-Wesly. Cambridge University Press.4 4. S.1 3. 3.4 2.3 2.3 References As most directly related books to these notes. Diagrammatica.V.2 2. Quantum Field Theory. An introduction to Quantum Field Theory. 2. Veltman.4 3. Schroeder.2 2. Quantum Field Theory.8 3. Cambridge University Press. McGraw-Hill. M.3.1 2.3 2.5 5.1 5. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. 3.1 4. Weinberg.3 3.2 2. 1. II: Modern Applications. The quantum theory of ﬁelds.4 3.2 6. Srednicki.5 2.1 2.H.4 2. 4.3 4.5 4. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012). 5.1 6. 3.4 3. Ryder.1 2 2 3. Corresponding chapters in books of Ryder. but are rather compact. Cambridge University Press. C. Cambridge University Press.2 36 22 .3 3.3 3. L. 1995.2 4. Itzykson and J.2 3. Quantum Field Theory. Zuber.2. 6. 1985.3 3. I refer to the book of Srednicki [1] and Ryder [2]. M. 1994. Vol.3 2.2 3. 2007. I: Foundations.2 6. The notes contain all essential information.1 3.2 3.

5 9.3 6.2. 6.3 2.2 9.6 3.2 12.7 4.6 6.8 6.3 6 8 9 Srednicki 3 3 3 37 5.1 20.8 5 4.1.6 4.4 12. 8. 20.6 8. 2. 2.2 10.1.4 3.4 4.5 6 7 6 6.4 9. 4.3 7.2 4.1 10. 4. 6.1 12. 6. 9.3.5 7.4 8.1 11.2 11.4.2. 9.1.5 7. 6.6 3.5 2.7.6 9.1 5.3 9.10 8.5 9.2.5 4.3 11.2.2 8.1 20.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.5 6.6 10.5 11.3.4 10.4 7.2 7.3 10.3 20. 4.4. 4.3 12.1 9.3 8.4 These notes Section 7.4.1 20.1 8.4 12.3 8.4 2.3.7 5.5 .1 9.2 8.3 3.6 3.5.2 5.1 7. 6.3 7.5 4.

• Slight rewriting of discussion of Majorana mass term in section 6.5 Corrections 2011/2012) • Extension of Exercise 3.2 .3 • First two sentences of Chapter 5 have been modiﬁed.

054 571 68 (18) × 10−34 J s = 6. (1. They satisfy the well-known (canonical) operator commutation relations [ri . S ≫ . ≡ h/2π = 1. the action of the momentum operator in the coordinate representation is not as simple as the position operator. c = 299 792 458 m s−1 . ˜ ψ(p) = i d3 r exp − p · r ψ(r).1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. In quantum mechanics a special role is played by Planck’s constant h. (1.1) In the limit that the action S is much larger than . Indeed. (1. (1.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). Corresponding position and momentum operators do not commute. In quantum mechanics.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. 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.Chapter 1 Introduction 1. quantum eﬀects do not play a role anymore and one is in the classical domain. One can talk about states |r and the wave function ψ(r) = r||ψ . pj ] = i δij . It is given by pop ψ(r) = −i ∇ψ(r). ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. usually given divided by 2π. In special relativity a special role is played by the velocity of light c. 1 (1. (1. the position can in principle be determined at any time with any accuracy. being eigenvalues of the position operators.6) .2) In the limit that v ≪ c one reaches the non-relativistic domain.4) where δij is the Kronecker δ function.582 119 15 (56) × 10−22 MeV s.

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.

x2 . 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 . and are given by (x0 . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). The Poincar´ transformations include e e Lorentz transformations and translations. Unlike in Euclidean space.. Because of the diﬀerent signs occurring in gµν . x2 = x · x = xµ xν nµ · nν = xµ xν gµν . When 3-dimensional space is considered as part of Minkowski space. x2 .26) (1. ˆ ˆ (1.21) (1. One has x2 = xµ xµ = t2 − x2 . we will use upper indices for the three-vectors. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . the points lie on the lightcone and they can be connected by a light signal. x) = (x0 .3). • ds2 < 0 (spacelike intervals). xµ = gµν xν . 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. it is convenient to distinguish lower indices from upper indices.23) δjm δkn − δjn δkm . Many other four vectors and tensors transforming like T ′ µ1 .. the invariant length or distance (squared) is not positive deﬁnite. Relations relating tensor expressions.νn can be conν structed. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 . (1.2. Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . −x).. No diﬀerence is made between upper or lower. (1. 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.. x1 . in which case zi = ǫijk xj yk . one must distinguish tensors with upper or lower indices and one can have mixed tensors. One can distinguish: • ds2 > 0 (timelike intervals).3) with basis nµ (µ = ˆ 0. linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. x3 ). • ds2 = 0 (lightlike or null intervals).Introduction 5 that can be used in the cross product of two vectors z = x × y. We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices.µn = Λµ1ν1 . The real. however. x1 . In special relativity we start with a four-dimensional real vector space E(1. xi are the three space components. These transformations do change the components of a vector. The lower indices are constructed in the following way. The (invariant) lengths often have special names. denoted as V ′µ = Λµν V ν .24) The quantity gµν ≡ nµ · nν is the metric tensor. are . In Minkowski space. where the coordinate t = x0 is referred to as the time component. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. δim δjm δkm δin δjn δkn .22) (1.1. in this case an inertial system exists in which the two points are at the same space point and in that frame ds2 just represents the time diﬀerence ds2 = dt2 . Λµn n T ν1 . The length (squared) of a vector is obtained from the scalar ˆ product.25) Within Minkowski space the real. So we could have used all upper indices in the above equations. x3 ) = (t. . . independent of a coordinate system. Vectors are denoted x = xµ nµ .

35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ .∇ .34) (1. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . ∂3 ) = ∂ ∂ ∂ ∂ . 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 . because one has (1. for instance. but more important has the same eﬀect on lefthandside and righthandside. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. (1. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. (1. Note that in this equation one has on left. 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. ∂2 . (1. The length squared of ∂ is the d’Alembertian operator.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. .and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector.36) Exercises Exercise 1. starting from ǫ0123 = 1. .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. (1. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . (1. explicitly written as (p0 . Each of these permutations leads to a minus sign. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . where m is called the mass of the system. It is an invariant tensor. a∞ = 4πǫ0 2 /me e2 . g0 = g1 = g2 = g3 = 1. . For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .27) It is easy to convince oneself of the nature of the indices in the above equation.32) .Introduction 6 called covariant. constructed via the metric tensor itself. not aﬀected by Lorentz transformations.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’.28) ν ν Note that gµ with one upper and lower index. ′ ′ . ∂1 .33) (1. = −24. p). ∂t (1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . p) = (E. deﬁned by ∂2 − ∇2 . = ∂ .

which is the Compton wavelength for the Planck mass. ν. . n1 . c. 1. 2. 0. a2 . a1 . aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. ˆ ˆ .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . a+ .3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . n1 . [Note: what is the MKS unit for V/T?] Exercise 1.(A · B) C using the properties of the tensor ǫijk given in section 1. 0. 0) ˆ n3 = (0. . ]. me .Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). 1. 0. Compare it with the proton mass and use Eq. 0.2 Prove the identity A × (B × C) = (A · C) B .13 to give its actual value in kg. by a simple few-line reasoning [For instance: If {µ. n2 . One does not need to know . 0) ˆ n2 = (0. n3 . 1. How are a+ and a− related to a+ and a− . re = e2 /4πǫ0 me c2 to the Compton wavelength. ρ. ˆ ˆ ˆ ˆ These are n1 = (0. a+ . . n2 . a2 ] or [a− . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . 1) ˆ . (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). This demonstrates how a careful use of units can save a lot of work. 3} the index α can only be equal to one of the indices in ǫµνρσ . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . a+ . 0) ˆ n0 = (1. Also for the coordinates [a− . e. 0. (b) The coordinates (a0 .3. a1 .511 MeV. a2 ] we can ﬁnd basis vectors n− . Also construct and calculate the Planck length Lpl .71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). (d) Use the value of the gravitational constant GN / c5 = 6. Exercise 1. n+ . σ} is a permutation of {0. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. but only appropriate combinations. 0. g−− . g+− and g−+ . Exercise 1. a3 ) are the expansion coeﬃcients using the basis vectors n0 . α and me c2 = 0. ǫ0 . 1. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . a1 .

namely plane waves characterized by a wave number k.4) where M is the particle mass.2. t) = exp(−i k 0 t + i k · r). Although it is straightforward to ﬁnd the solutions of this equation.3) Equations 2. ∂t2 (2. 2 + M 2 φ(r. t) = 0. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ.1 The Klein-Gordon equation In this chapter.2) Acting on the wave function one ﬁnds for a free particle.3 are not covariant. φk (r.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . t) = ∂2 − ∇2 + M 2 φ(r. (2.7) k 0 = ± k2 + M 2 = ±Ek . (2.Chapter 2 Relativistic wave equations 2.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. 8 . 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. i ∂ ∇2 ψ(r. ∂t 2M (2. 2M (2. p). ∂t p −→ pop = −i∇. But the replacement 2. t) = − ψ(r. written as pµ −→ i∂µ is covariant (the same in every frame of reference).5) with (k 0 )2 = k2 + M 2 . The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.1 and 2. E = p2 . p2 = pµ pµ = E 2 − p2 = M 2 . Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. (2. one obtains (2. t).

As we will see later both problems are related and have to do with the existence of particles and antiparticles.5). φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . This leads to the existence of negative densities.10) ∂t which follows directly from the Schr¨dinger equation. 2M 2M (2. (2. there are no longer fundamental objections to mix up space and time. This continuity equation can be written down o covariantly using the components (ρ. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. ∂ρ = −∇ · j. k) ≡ a(k) and φ(−Ek .11) Therefore. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. The KG equation. relativistically the density is not a scalar quantity. for which we need the interpretation of φ itself as an operator.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. −k) . k) + ei k·x φ(−Ek . 2.9) They satisfy the continuity equation. The appropriate current corresponding to the KG equation (see Excercise 2. (2. ∂µ j µ = 0.14) ˜ ˜ Introducing φ(Ek . j) of the four-current j.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. And. there are solutions with negative energy. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. −i φ∗ ∇ φ . 2. but rather the zero component of a four vector.8) (2. rather than as a wave function. just as the dependence on time was in ordinary quantum mechanics. j) = ↔ i φ∗ ∂0 φ. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. having the KG equation as a space-time symmetric (classical) equation.15) . is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. Then. however. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends.Relativistic wave equations 9 i. which is what Lorentz transformations do. ↔ ↔ (2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). This operator has all possible solutions in it multiplied with creation (and annihilation) operators. 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. (2.e.3). we want the most general solution.13) ˜ with (in principle complex) coeﬃcients φ(k). (2π)3 2Ek (2.2) is j µ = i φ∗ ∂ µ φ or (ρ.

(2π)3 2Ek (2. (2. (2.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . which implies (xµ ) = (t.6 We will explicitly discuss the example of a discrete symmetry. x) → (t.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). x ∂µ ∂ µ + M 2 φ(˜) = 0. −x) ≡ (˜µ ). It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. φP (x) ≡ φ(˜) (P for parity).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.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). i. after restricting the energies to be positive). This shows how parity transforms k-modes into −k modes.23) . (2π)3 2Ek (2. Since a · b = a · ˜ it is easy to see that ˜ b. 2.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy.18) (2. Another symmetry is found by complex conjugating the KG equation. 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) . The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.e. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). 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) . x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). for which we consider space inversion. changing the sign of the spatial coordinates. Performing some Lorentz transformation x → x′ = Λx. (2π)3 2Ek (2.15 one has elimated k 0 and in both equations k · x = Ek t − k · x.16) The consequence of this is discussed in Exercise 2.14 and 2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes).17) (2.Relativistic wave equations 10 In Eqs 2. (2.

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

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

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

0 (3. They satisfy the commutation relations [Li . ˆ therefore. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). With matrix commutation this algebra satisﬁes the requirements for a Lie algebra. [z. π) (n. y] ∈ A. One way of viewing the parameter space. These generators form a threedimensional Lie algebra SO(3). we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. Ly and Lz .1: the parameter spaces of SO(3) (left) and SU (2) (right). Locally this is a 3-dimensional Euclidean space and SU (2).z N N or (Lk )ij = −i ǫijk . y ∈ A ⇒ [x.6) Summarizing. y]] = 0 (Jacobi identity).4) R . x]] + [z. with real a’s and 3 (ai )2 = 1.and y-axes 0 0 0 0 0 0 0 −i . y] = −[y. z ) = lim N →∞ that are generated by i 0 . namely that there exists a bilinear product [.Groups and their representations 14 (n. Lj ] = i ǫijk Lk . ˆ R(ϕ. (3. x]). 2 (3. (3.5) Rotations in general do not commute. is a 3-dimensional Lie-group. In the same way we can consider rotations around the x. which reﬂects itself in the noncommutation of the generators. Next. Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. x] = 0 (thus [x. [y.2 π) (-n.7) = a0 0 −1 i 0 1 0 0 1 • [x. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. [x. Lx = Ly = 0 0 0 i 0 −i 0 ϕ ˆ . the rotations in SO(3) can be generated from inﬁnitesimal rotations that can be expressed in terms of a basis of three generators Lx . It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. . π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . ] that satisﬁes • ∀ x. • [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. z]] + [y. and thus (σ · n)2 = 1.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). for rotations around z for instance.

Near the identity. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. H] = 0 for g ∈ G. (3. i. . (3. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. . thus. it is suﬃcient that the generators commute with H. For instance in a function space translations. .9) The parameter-space of SU (2). thus. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. 2! . σy /2 and σz /2 form the basis of the Lie-algebra SU (2). ˆ ˆ A(ϕ. immediately sees that the Lie algebras are identical. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). . The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. also can be considered as a ﬁlled 3-sphere. i. n) ˆ −→ R(2π − ϕ. the above mapping corresponds to the trivial mapping of the Lie algebras.Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. 2 2 2 (3. i. n) π ≤ ϕ ≤ 2π.14) For a Lie group.15) [g.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. (3.e. . n! 2! n=0 ∞ (3.12) One. n) −→ R(ϕ. however. . The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. They satisfy σi σj σk = i ǫijk . . sometimes in more than one way (but this will be discussed later).e. [g. 3. φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . in particular that AB → RAB = RA RB ). 2 (3.11) Thus σx /2. n) ≈ 1 + i ϕ J · n. for ﬁxed n. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ.10) σ ˆ · n.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. H] = 0 for g ∈ G. In the full parameter space. Suppose we have a system described by a Hamiltonian H. n) i ∂ϕ = ϕ=0 (3. SU (2) ≃ SO(3).e.

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

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

consisting of the Lorentz group and translations. it is convenient to use a vector notation for the points in Minkowski space. 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. From this property. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .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 (3.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. We considered the generators and looked for representations in ﬁnite dimensional spaces. det Λ = −1 is called improper). it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.22) (Note that xT is a row vector). e To derive some of the properties of the Lorentz group. (ii) |Λ00 | ≥ 1. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = . Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. σ/2 in a two-dimensional (spin 1/2) case. (3. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ .26) (3. Invariance of the length of a vector requires for the Lorentz transformations x′2 = gµν x′µ x′ν = gµν Λµρ xρ Λνσ xσ = x2 = gρσ xρ xσ or Λµρ gµν Λνσ = gρσ . In this section we consider the Poincar´ group.25) (3. G = 10 (3.Groups and their representations 18 3.g. (det Λ = +1 is called proper.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . Writing x as a column vector and the metric tensor in matrix form.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. e.27) (3.

J j ] = i ǫijk J k . βγ = sinh η.32) 0 − 0 −1 0 0 0 0 −1 . J 3 ). Returning to the global group. This is the origin of the Thomas precession. K j ] = −i ǫijk J k . Using these explicit transformations. 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. J = (J 1 . 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. inﬁnitesimally given by ΛB (η. In L↑ . (3. Thus ˆ 0′ 0 0 0 V 0 V 1 0 0 0 0 1′ 1 V 0 cos ϕ sin ϕ 0 V 0 0 −i 0 2′ = 2 −→ J3 = V 0 − sin ϕ cos ϕ 0 V 0 i 0 0 . 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. (3. the diﬀerence of ﬁrst performing a boost in the y-direction and thereafter in the x-direction and the boosts in reversed order) contains a rotation (in the example around the z-axis).29) 0 0 1 0 3′ 3 V sinh η 0 0 cosh η V i 0 0 0 Note that Lorentz transformations generated from the identity must belong to L↑ .Groups and their representations = i=1 (Λ0 )2 . Rotations around the z-axis are given by ΛR (ϕ. From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. K 3 ). K j ] = i ǫijk K k . which satisfy the commutation relations (check!) [J i . J 2 . Thus 0′ 0 V cosh η 0 0 sinh η V 0 0 0 i 1′ 1 V V 0 0 0 0 0 1 0 0 2′ = 2 −→ K3 = V V 0 0 0 0 . [J i . i proof: use ΛT GΛ = G and ΛGΛT = G. z ) = exp(i ϕ J 3 ). we have found the generators of rotations. inﬁnitesimally given by ˆ ΛR (ϕ.g. K 2 . z ) = exp(−i ηK 3 ). and those of the boosts.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. it is easy to ﬁnd the following typical examples from each of the four classes.28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. z ) ≈ ˆ ˆ I − i η K 3 . K = (K 1 . since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. [K i . z ) ≈ I + i ϕ J 3 . The commutator of two boosts in diﬀerent directions (e. one distinguishes rotations 0 + and boosts.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations.

J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. (3. 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 ) . [M µν . Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. we continue using covariance arguments. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). We therefore ﬁnd (again) that there are six generators for the Lorentz group. 3 30 with η = ω and K = M (e. i i0 i i0 1 ijk M jk 2ǫ (3. There are six generators. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 .g. Summarizing. inﬁnitesimally approximated by (I + ω.39) (e. We will just require Poincar´ invariance for the generators themselves. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3.36) (3.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. This is a + normal subgroup and the factor group is the Vierer group. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 . One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν .g.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .33) These four transformations form the Vierer group (group of Klein). (3.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0.38) Explicitly. Of the four parts only L↑ forms a group. which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . ǫ). a).35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . 3. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. K = M ).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. P+ etc. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . inf (3. The extension to the Poincar´ group is e ↑ straightforward. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . three of which only involve spatial coordinates (rotations) and three others involving time components (boosts). In order to ﬁnd the commutation relations including the translation generators.

46) [K i . 3. because one then sees that by letting c → ∞ the commutation relations of the Galilei group. a) = Λ−1 µ P ν . H] = 0. a)P µ U (Λ.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. writing the generator P = (H/c. They just state that M µν transforms as a tensor with two Lorentz indices.47) (3. The generators of the translations are the (momentum) operators P µ . Explicitly. Taking any one of the states in an irreducible representation. 2 2 (3. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . K j ] = −i ǫijk J k /c2 . a)P µ U † (Λ. known from non-relativistic quantum mechanics are obtained.44) (3. . check that U1 U2 corresponds with Λ1 Λ2 ). H] = i P i .41) 2 At this point.45) Note that the commutation relations among the generators M µν in Eq. H] = [J i . We have here reinstated c.38 could have been obtained in the same way. = −i (g µρ P ν − g νρ P µ ) .43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . ǫ) = 1 − ωαβ M αβ + i ǫα P α . Inﬁnitesimally. (3. P j ] = i ǫijk P k . a) = Λµν P ν or U † (Λ. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . [K i . P ρ ] = 0. the generators J 2 and J 3 in the case of the rotation group. (3. Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . . The eigenvalues of these will be the four-momentum pµ of the state. . leads to U (Λ. .48) . the generators M αβ no longer exclusively act in Minkowski space. [J i . one obtains 2 [P i . (3. the generators J± in the case of the rotation group. P ν ] [M µν . J j ] = i ǫijk J k . . [J i . . For instance. To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators. These form a group called the little group associated with that four vector. [K i . For instance.42) (to decide on where the inverse is. P j ] = [P i . [P µ . P j ] = i δ ij H/c2 . = pµ |p. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). ν (3. that commute e among themselves. P µ |p. other states in that representation are obtained by the action of operators outside the maximal commuting set. . (3. In order to label the states in an irreducible representation we construct a maximal set of commuting operators.5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. as part of the set. P ) in terms of the Hamiltonian and the three-momentum operators. [J i . 3. K j ] = i ǫijk K k .

(3.52) (3. 2 (3.51) The following properties follow from the fact that W µ is a four-vector by construction.54) This will enable us to pick a suitable commuting ’spin’ operator. Wν ] = −i (gµα Wν − gνα Wµ ) . Massive particles: p2 = M 2 > 0. From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . = = i exp − ǫµνρσ M µν pρ sσ |p.55) commute with all generators and therefore are invariants under Poincar´ transformations.53) (3. . . which can thus be chosen spacelike. . 2 exp (−isµ W µ ) |p. 0.49) (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. M ǫijk . . p0 > 0. The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. . Pν ] = [Wµ . . while the others have no physical signiﬁcance. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . Wα ] = [Wµ . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . 0. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . . the third case represents the vacuum.56) .Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). i ǫµνρσ W ρ P σ . (3. .

(3. Together with the four momentum states thus can be labeled as |M. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. W i (p) = −W · ni (p)).62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0..57 The best insight in the meaning of the operators W µ is to sit in the rest frame of the particle and put P µ = (M. s. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). 2 . M2 (3. p0 > 0. In that case the vectors ni (p) are just the space directions and W = (0. 0. M J) with components W i = (M/2)ǫijk M jk .60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . n1 (p). 0.e. Having made a covariant decomposition. (3. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). p.65) . 1. 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 . |p3 |). 1. M2 (3. 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. 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 = 0.59) (3. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. 0).64) such that in an arbitrary frame where the four vectors p.61) one sees that S i (p) i =− W2 . 0).57) (i. 0. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. . (1.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). 1. show that W /M can be interpreted as the intrinsic spin. W j (p)] = i M ǫijk W k (p). ms . . 3. i (3. 0).e. (0. 0. 1 2 (3. 0.58) i.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. (0. The commutation relations [W i (p). 0. . S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p).

67 has a vanishing right hand side). however. 3. Note that angular momentum does not contribute to helicity as L · p = 0.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations.67) (3. the action of RA b · RA a = = ˆ ˆ ˆ b · a cos ϕ + n · (b × a) sin ϕ + 2(b · n)(a · n) sin2 (ϕ/2) ˆ ˆ b · a + b · (ˆ × n) sin ϕ − 2 (b · a − (b · n)(a · n)) sin2 (ϕ/2) a . (3.73) (3. 3. (3. and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0). Thus 2 λ(p) = J ·p . Wν ]. A massless particle. The algebra of W 1 . P µ .1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises).Groups and their representations The above does actually include the massive case. i W 2 (p). U (Λ. The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). W 2 (p)] = = = i M 2 λ(p).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p).66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). |p| (3. W 2 (p)] [λ(p).68) (3.13. Thus we have P 2 . a)|0 = |0 . We don’t know of any physical relevance. 1 2 24 (3. and gives [W 1 (p).70) as a commuting set with zero eigenvalues for P 2 and W 2 . W 2 (3. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). Exercises Exercise 3. W µ (p) = λ(p) P µ . thus. λ . 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. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. W 2 and λ is derived from the general commutation relations for [Wµ . But if M = 0 one has a diﬀerent algebra (Eq. is characterized by its momentum and the helicity. |p. which can take any integer or half-integer value. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). W 1 (p)] [λ(p). This state is in physical applications nondegenerate and is denoted by |0 . The vacuum: pµ = 0. −i W 1 (p).71) which is called the helicity. implying continuous values for the spin (actually for W i = M S i ).

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

6 Consider rotations and translations in a plane (two-dimensional Euclidean space).e. = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. you might just check relations involving J or K by taking some (relevant) explicit components.Groups and their representations 26 Exercise 3. although the presence of an interaction term V (r1 . that’s why many-particle non-relativistic quantum mechanics is ’easy’.67 to 3. r2 ) inevitably leads to interaction terms in the boost operators. [ri . momentum and spin operators e satisfy the canonical commutation relations. derive the commutation relations and show that the algebra is isomorphic to (i. i. i. if you don’t want to do this in general.69 of the Pauli-Lubanski operators for massless particles. they describe r −→ r′ = r − u t while t′ = t. show ﬁrst the operator identity [r. sj ] = iǫijk sk . Hint: for the Hamiltonian. rj ] = [pi . Comment: extending this to more particles is a highly non-trivial procedure.7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. H] + i dt ∂O . consistent with the (canonical) commutation relations of a quantum theory. sj ] = 0.5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame.e. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O. boost invariance leads to a conserved quantity. This is possible for a single particle. the others vanish. f (p)] = i ∇p f (p). pj ] = iδ ij and [si . although that in this case is a relation that is found already from the Ehrenfest relation for O = ri .e. Exercise 3. .. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. there is a one to one mapping) the algebra in Eqs 3. Do this by showing that the set of operators. These do not matter in the non-relativistic limit (c → ∞). H P J K = p 2 c 2 + m2 c4 . For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). indeed assures d K /dt = 0. but it can be done. [ri . Determine the commutation relations for the corresponding operators Ki in quantum mechanics. in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. pj ] = [ri . and U (u) p U −1 (u) = p − m u. (c) Give the unitary operator for boosts. Exercise 3. (a) In classical mechanics. From either of these. = r × p + s. p×s 1 (rH + Hr) − t p + . sj ] = [pi . = p. the operator U (u) that gives U (u) r U −1 (u) = r − u t.

9) (4. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. Special cases are the following representations: Type I : (j. K = i J (A = 0). j) K = −i J (B = 0). They will be labeled by two angular momenta corresponding to the A and B groups. U (ϕ) H(η) (4.3) (4. (4.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. It is simply connected and forms the covering group of L↑ . 2 1 (J − i K). where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. 0) Type II : (0.5) = 1 (J + i K). 2 2 σ σ J = . C) is the group of complex 2 × 2 matrices with determinant one. The matrices in SL(2. 2 2 (4. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . C) can be written as a product of a unitary + matrix U and a hermitean matrix H. They precisely correspond to the two types of representations that we have seen before. This tells us how to ﬁnd the representations of the group. (4.6) (4. The group SL(2. j ′ ).Chapter 4 The Dirac equation 4. B j ] = [Ai .10) .1) (4. Aj ] = [B i . 0. 2 (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 . becoming the deﬁning representations of SL(2. With this choice of parameter-spaces the plus and minus signs are actually relevant. C). K = −i . B j ] = i ǫijk Ak . K = +i .4) (4. i ǫijk B k . respectively.7) From the considerations above. similarly as the homeomorphism between SO(3) and SU (2). (j.1 The Lorentz group and SL(2.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ .

14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. (4. −σ). thus. ξ −→ Hξ. cannot be connected by a Lorentz transformation belonging to L↑ . (4. This shows that M ∗ ≃ M . U σ µ aµ U −1 = σ µ ΛR aµ . One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 .16) = σ µ Λ R aµ . U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. C).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. Λ(ǫ) ∈ L+ . One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2.18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. η → U η. Eqs 4. (4. The following relations for + rotations and boosts are useful. 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. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.15) (4. while ±ǫη ∗ transforms as a type-I + spinor.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. (ǫ∗ = ǫ. (4. σ µ ≡ (1. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1.11) Considering ξ and η as spin states in the rest-frame.12) ξ → U ξ. In fact. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. if there does not exist a unitary matrix S. ˜ (4. ˜ = σ µ Λ B aµ . such that M = SM S −1 . Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ).18) As already discussed for SU (2). however. The Lorentz transformations Λ(M ) and Λ(M ). ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. H ≡ (H † )−1 = H −1 ). namely H(η) = exp(η · σ/2). (4. C) or L↑ . Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . H(η) = exp(−η · σ/2). 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). ǫ−1 = ǫ† = ǫT = −ǫ). one can use a boost to the frame with momentum ˆ ˆ p.13) 2M (E + M ) (exercise 4. σ).1). i. thus. there exists the matrix ǫ = iσ 2 . η → Hη.19) .e. 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. C) and. This is not true for the two-component spinors in SL(2. C).

1. m) = χm uL H(p) 0 . −→ −p (vector). 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. We also have seen that the representations (0. a) and ˜ a = (a0 . they cannot be connected by a unitary 2 2 transformation. Taking M (Λ) = H(p). A parity transformation changes aµ into aµ . It has the same eﬀect as lowering the indices. For a particle at rest only angular momentum is important and we can choose ξ(0. (vector).2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. −a). (4. as we have seen in section 4.13. ǫµνρσ will change sign. −→ −r (vector). 2 ) and ( 2 . C).The Dirac equation 29 where The matrix I2 does not belong to L↑ . 0) of SL(2. but to L↑ .21) In nature parity turns (often) out to be a good quantum number for elementary particle states. generators. 0).22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq.e. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . we have seen that in the rest frame W i (p)/M → J i .17. The behavior is the same for classical quantities. therefore. (axial vector). (pseudoscalar). e.23) u(p. the aij elements of a tensor will not change sign. 1 ) and ( 1 . Although these operators cannot be used to label the representations. where a = (a0 . 2 2 (4. This provides the easiest way of seeing ˜ what is happening. 0 χm uR H(p) . i. 4. The representing member of the class P− is the parity or space inversion operator Is . must combine the repe resentations. etc. but rather the operators W i (p) should be used. (scalar). we obtain for the two components of u which we will refer to as chiral right and chiral left components. 4. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . 0) are inequivalent. Within the Lorentz group.g. 1 ) and ( 1 . From the deﬁnition of the representations (0. they can be connected. C) are suitable for representing spin 1/2 particles. −→ M (Λ) (4. the boost in Eq.20) 4. but by a transformation belonging ↑ ↑ to the class P− .e. ) −→ ( . that have the correct commutation relations. m) = χm . the well-known two-component spinor for a spin 1/2 particle. = (4. m) = η(0. The i angular momentum operators J are represented by σ i /2. Thus M ∗ and M are not equivalent within SL(2. For the spin 1/2 representations of the Poincar´ group including parity we. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. i. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar).

this can be rewritten 1 µ ν {γ . It is of a form that we also played with in Exercise 2. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.29) which is an explicit realization of the (momentum space) Dirac equation. (4. (4.28) which is a covariant (linear) ﬁrst order diﬀerential equation. 0 H 2 (p) uR uR . p (4. ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0.26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0. (4.33) deﬁne for a four vector a the contraction a = aµ γµ / . The explicit realization appearing in Eq. We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. From this one obtains the Cliﬀord algebra for the Dirac matrices. γ ν } = 2 g µν . (4. which in general is a linear equation in pµ . The general requirements for the γ matrices are thus easily obtained. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. (4. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.7.31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). γ } ∂µ ∂ν + M 2 ψ(x) = 0. = H −1 (p). . 2 (4.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. Applying (iγ µ ∂µ + M ) from the left gives (4. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ). We will discuss another explicit realization of this algebra in the next section.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . The ﬁrst indeed is solved.27 is known as the Weyl representation. = (4.24) (4. (iγ µ ∂µ − M ) ψ(x) = 0. Since ∂µ ∂ν is symmetric. 4. {γ µ .32) suppressing on the RHS the identity matrix in Dirac space.

γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. E = +M (twice) and E = −M (twice). 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. but with properties such that it can be annihilated by the particle (e.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0. 4. The Dirac sea forms the vacuum. We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.1: The Dirac sea of negative energy states and antiparticles. This hole forms an antiparticle with the same mass. (4. opposite charge). The density j 0 = ψγ 0 ψ = ψ † ψ (4.38) The general algebra for the Dirac matrices is Two often used explicit representations are the following2 : The standard representation: σk 1 0 0 0 3 k 2 k .35) is indeed positive and j 0 can serve as a probability density. . 4. .37) Using the explicit form of γ in the Weyl representation (see Eq. The second problem encountered before.36) (4. From the vacuum a particle can be removed. (4.3 General representations of γ matrices and Dirac spinors {γ µ .39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. remains. (4. Relying on the Pauli exclusion principle for spin 1/2 particles.g. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.27). one sees that there are two positive and two negative eigenvalues. the one of the negative energy states. γ ν } = 2 g µν . This can most easily be seen in the particle rest frame.34) for the adjoint spinor ψ ≡ ψ † γ0 . This problem was solved by Dirac through the introduction of a negative energy sea.

For example. −→ Sγµ S −1 . = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . γ 0 ].R.46) (4. 0 (4.43) S=√ . γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. = S(γµ )W.R. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4.9 and 4.R. 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 . 2 (4.R. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 .48) are projection operators.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 . while L−1 γ µ L = Λµ γ ν . the rotation and boost generators in Weyl representation in Eqs 4. which in the case of the Weyl representation are just the upper and lower components. that project out chiral right/left states. γ 2 ]. is 1 1 1 (4. (4. −→ Sψ.45) (γ5 )W.40) (4. γ σ ]. =ρ ⊗1= 0 −1 3 (4. (γµ )S. For instance in the Weyl representation (but valid generally). S −1 . 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 .41) (4. (γ5 )S. γ µ } = 0 and explicitly one has 0 1 1 .47) (4.3) . 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . We note that ψ → Lψ and ψ → ψ L−1 . 4! (4.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ .The Dirac equation 32 The Weyl (or chiral) representation: 0 1 .44) and an adjoint spinor deﬁned by ψ = ψ † γ0 . Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.

55) (4. (C)S.R. 0 −ǫ 0 0 −ǫ 0 . ψ= (4. One has ψ c = −ψ T C −1 .R. but it does. Some properties of C are C −1 = C † and C T = −C. −iσ 2 0 −ǫ = . which is again a solution of the Dirac equation. In S.57) (4. (i/ − M ) ψ(x) = 0. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. parity. x) → x = (t. The latter implies that the conjugate of a right-handed spinor. Explicitly. ψR . This symmetry does not change the spin 1/2 nature. e. called charge conjugation. ∂ (4. T (4. C is real and one has C −1 = C † = C T = −C c and [C.56) (i/ − M ) ψ c (x) = 0. = = 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. for instance. (C)W. As with parity we look for a transformation. such that T Cγµ C −1 = −γµ .54) e.g. ∂ (4. γ5 ] = 0.52) η ξ one can apply space inversion. In any representation a matrix C exist.g.53) (4. (or W. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η .51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. ∂ with the solution ψ c (x) = ηc Cψ (x). T (4.R. reverse the charge of the particle. that brings ψ → ψ c . x = (t. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. is a left-handed spinor.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.) and all representations connected via a real (up to an overall phase) matrix S.59) −→ ψ c = C ψ = η ǫ ξ∗ .The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ .58) where ηc is an arbitrary (unobservable) phase. 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. ψ= (4. usually to be taken unity.6). = 0 ǫ iσ 2 (4.R.

The ¯ ¯ solutions are normalized to u(p. ¯ ¯ † ′ u (p. s)v(p. p (4. ¯ ′ ′ Explicit solutions in the standard representation are χs . s).65) (4.64) (4. s)u(p. s ) = v (p. s ) = 0. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4.The Dirac equation 34 4. s) + v(k. 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. s)u(p.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. ⇔ (/ − M ) u(p) = 0. s′ ) = 2M δss′ . ⇔ (/ + M ) v(p) = 0. v(p. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. where u satisﬁes −σ · p Ep − M u(p) = 0. 2M 2M −/ + M p v(p.68) (4. ∂ −iσ · ∇ −i ∂t − M (4. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. v (p. s)¯(p.63) u(p. ψ(x) = v ±s (p) ei p·x .69) = − s In order to project out spin states.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. In that frame is a spacelike unit vector sµ = (0. with E = Ep = + p2 + M 2 . s)u(p. s) u = . s) = v(p. s) = u(p. in which γ 0 is diagonal (see discussion of negative energy states in section 4. s)v(p. s′ ) = 2Ep δss′ . the spin polarization vector in the rest frame is the starting ˆ point. s) ei k·x d∗ (k. s).70) = Ps = ˆ 0 1∓σ·s 2 2 . the spinors satisfy C uT (p. s) . be obtained by a Lorentz transformation. (2π)3 2Ek (4. s)¯(p. s) and C v T (p.62) where χs are two independent (s = ±) two-component spinors.g. s ) = v † (p. (4.61) σ·p −(Ep + M ) There are also two negative energy solutions. ψ(x) = u±s (p) e−i p·x . s) e−i k·x b(k.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. In an arbitrary frame one has s · p = 0 and s(p) can e. s′ ) = −2M δss′ . s) = Ep + M σ ·p Ep +M χs . where v satisﬁes σ·p −(Ep + M ) v(p) = 0. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s) v = . the best representation to describe particles at rest is the standard representation. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation).60) ψ(x) = 0. s)v(p. ¯ (4. 2M 2M (4..1). s) = u(p.

p ψR/L ≡ PR/L ψ are independent solutions. λ = −1/2) = √ − E+M √ 0 0 E+M (4. u(p. 4. λ = − 1 ) = √ E + M − E − M . −p3 −p1 − i p2 1 v(p. λ = −1/2) = u(p. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . +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 √ . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. In principle massless fermions can be described by twocomponent spinors. +1/2) = √ u(p. Explicit examples of spinors are useful to illustrate spin eigenstates. (4.74) . λ = + 1 ) = √ + E +M − E −M . λ = − 1 ) = √ √ v(p. etc. λ). For instance with the z-axis as spin quantization axis. √ 0 u(p. .73) Also for helicity states C uT (p. Therefore. λ) and C uT (p. −1/2) = √ (4. v(p. λ = +1/2) = v(p. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. chirality.71) p − i p2 .48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). one has in standard representation: 0 E+M E+M 1 1 0 . while the negative helicity (λ = −1/2) is in essence lefthanded. λ) = u(p. 1 u(p. γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. u(p. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). helicity states.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . (4. This means that in the solution space for massless fermions the chirality states. 2 2 0 √ 2 √ 2 − E+M − E−M 0 . λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case.The Dirac equation 35 (the last equality in the restframe and in standard representation) projects out spin ±1/2 states (check ˆ ˆ this in the restframe for s = z ). v(p. The chirality projection operators in Eq. Note that for solutions of the massless Dirac equation /ψ = 0. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . λ) = v(p. −1/2) = √ E + M −(E + M ) 0 .

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

s). where x = (t.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. x ˜ . x → x. s)¯(p. 4. Excercise 4.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . s) and v(p. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n.6 Apply space-inversion. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.1 Prove Eq. 0 if n is odd. Exercise 4. H(p) = exp η·σ 2 = M +E+σ·p . s) u p /+M = 2M 2M both sides are projection operators. of which only the traces of four and two gamma-matrices are nonzero.g. Show that the equality holds.g. Do this e.13. a 4 a · b. 4 2 Exercise 4. (4. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν . −x) is also a solution of the Dirac equation. γ σ ] = σ ρσ . to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). e. by letting them act on the four basis spinors u(p. 1 i −i S ρσ = − [γ ρ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M.77) Exercises Excercise 4. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ .5 Show starting from the relation {γµ . ˜ Exercise 4.4 Show that in P+ ≡ s=±1/2 u(p.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4.

Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. Do you understand why? The calculation of the life-time is completed in chapter 11.The Dirac equation 38 Exercise 4.57 MeV. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e . me = 0. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. there is no current ﬂowing through the surface of the cavity. For such a cavity nµ = (0. . Excercise 4. n It turns out that there is actually a whole family of conditions that provide conﬁnement. mµ = 105.8 A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino (¯ℓ ). Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. Calculate for now just |M |2 . r).511 MeV). (b) Next.8 × 10−6 . namely i / ψ = eiαγ5 ψ. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. The pion has spin 0. i. Note: To prove this. ν p Note: how such parametrizations work will be explained in chapter 8.e. we want to conﬁne this solution to a spherical cavity with radius R. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). 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. you need to determine the corresponding conﬁnement condition for ψ. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). Note: use ﬁrst the appropriate equation (which?) that is obeyed by each individual component of the Dirac equation to show that the upper two components are allowed solutions. M R = 1. is a stationary solution of the (free) Dirac equation. Calculate the value of k for M R = 0. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R.66 MeV. the decay would even be forbidden. We will not pursue this chiral freedom at this point and take α = 0.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x.5 and M → ∞.

In the restframe.Chapter 5 Vector ﬁelds and Maxwell equations 5. there are then indeed three independent possibilities.± 5. A).1 Fields for spin 1 In section 3. (2 + M 2 )Vµ = 0. In an arbitrary frame ǫµ (k) with λ = 0. The vectorﬁeld is therefore (λ) suited to describe spin 1. (5.2 The electromagnetic ﬁeld F µν = = ∂ µ Aν − ∂ ν Aµ 1 −E 2 0 −E 1 E 0 −B 3 2 E 3 B 0 3 E −B 2 B 1 Aµ = (φ. These are solutions of the Klein-Gordon equation with in addition a condition. namely ǫµ (k0 ) = (0. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0.4) λ=0.3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. with ∂µ V µ = 0 (Lorentz condition). just as the two-component Dirac spinors did for spin 1/2. ±(1) are obtained by boosting the restframe vectors.1) The Lorentz condition implies that k µ ǫµ (k) = 0.2) (5.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. 39 (5. M2 (5. These can describe spin 1 states in the rest-frame. ǫλ ). where k = k0 = (M.5) −E B2 . λ) + ǫ(λ)∗ (k) eik·x c∗ (k. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x .2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation.7) . µ (2π)3 2E µ (5. λ) . one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν .6) where Aµ is the four-vector potential (5. 0).

10) the electric and magnetic ﬁelds are unchanged.17) (radiation. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν .16) (5. . moree over. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. 2Aµ = 0. AP (x) = Aµ (˜). j) the Maxwell equations read ∂µ F µν = j ν .15) (5. Therefore if k µ = (|k|. ∓1. it will vanish at all times. (5.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. 0. (5. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . transverse or Coulomb gauge). |k|). shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. Furthermore.11) This equation is not aﬀected by a gauge transformation. As remarked above. ∂µ Aµ = 0. 0).Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ.12) (5. 5. if ∂µ Aµ = 0 for large times |t|. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . Although the Lorenz condition is a constraint. It states the absence of magnetic charges and Faraday’s law. in accordance with what we found in section 3. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). (5. one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. (5. The equation in vacuum. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. Taking the divergence of this equiation. i. ˜ x Note that via parity transformation one obtains another solution. only two independent vectors remain ǫ(±) µ (k) = (0.8) (5.15 the d’Alembertian is replaced by 2 + M 2 .13) Hence. One possible choice is the gauge A0 = 0 (5. if in Eq. 40 (5. (5. −i.14) of which the solutions give the Li´nard-Wiechert potentials. 0.e. in which case one has 2Aµ = jµ . For electrodynamics one has furthermore the freedom of gauge transformations. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. for M = 0. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld.5.

Inside solenoid: Outside solenoid: B = B z. may look a bit akward. = 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. .0 . A= By Bx Br ϕ= − ˆ . sin ϕ. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ.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.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. This occurs in the region outside the solenoid. It is however. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). The situation. while the electromagnetic ﬁeld Aµ is not. 0) and ϕ = (− sin ϕ. cos ϕ. 2 2 2 BR2 BR2 y BR2 x A= .1: The Aharonov-Bohm experiment 5.0 .1. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . λ L − λ λ or ∆x = (L− λ/d)δ. F = e E + ev × B (5. illustrated in ﬁg. (5. 5. . 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 . . nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). To be precise. ˆ B = 0.Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). nevertheless. This causes an interference pattern as a function of ∆x. 0). but where E = B = 0. This phase diﬀerence is actually observed. where Aµ = 0.19) are gauge invariant. but A = 0 (hence there is structure in the vacuum). Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment.

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

∂t ∂B = 0. ∂t ∇ · B = 0. 5.8 and 5. .Vector ﬁelds and Maxwell equations 43 Exercises Exercise 5.9 explicitly give the equations ∇ · E = ρ.1 Show that Eqs. ∇×B =j+ ∇×E+ ∂E .

thus t′ = t + δτ.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. (6. ∂x ˙ (6. As an example. ˙ 2 (6. ˙ p= ∂L .8) 44 .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 .7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. t1 (6. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. recall classical mechanics with the action t2 S= t1 dt L(x. x) = K − V = ˙ 1 mx2 − V (x). x) ˙ (6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). ∂x (6.1) and as an example the lagrangian L(x. d dt ∂L ∂x ˙ = ∂L .Chapter 6 Classical lagrangian ﬁeld theory 6. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.5) with ∆x(t) = δx(t) + x(t) δτ . and the total change x′ (t′ ) = x(t) + ∆x(t). These equations can be obtained from a lagrangian using the action principle.3) (6.6) The ﬁrst term leads to the Euler-Lagrange equations. x (t) = x(t) + δx(t).

− d3 x . L (φ(x). . . ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ . . The hamiltonian H is deﬁned by H(p. H is a conserved quantity. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). λ=0 (6. In general any continuous transformation. x).12) Here R indicates a space-time volume bounded by (R3 . = ∂R (R3 . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . ∂µ φ′ . x′ ) − d4 x L (φ.11) λ=0 of which p and H are the simplest examples related to space and time translations.9) which is done because the ﬁrst term (multiplying ∆x. with ∆φ(x) = δφ(x) + (∂µ φ)δx . In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). t2 ).18) 1 Taking a functional derivative. think for instance of a temperature or density distribution or of an electromagnetic ﬁeld).t1 ) (6. ∂µ φ. t1 ) and R3 .6 can be rewritten as δS = . ∂µ φ(x)) = R d4 x L (φ(x). .e. 6.14) (6. + ∂L ∆x − H δτ ∂x ˙ t2 . indicated with δF [φ]/δφ should pose no problems. .15) d4 x′ L (φ′ . but which in 3 dimensions includes conserved quantities related to rotations and boosts. . = φ(x) + δφ(x) (6. (R3 . We will come back to it in a bit more formal way in section 9. δ(∂µ φ) (6. . x) ≡ p x − L .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ .Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). ∂µ φ(x). ∂µ φ(x)).17) Furthermore the variations δφ and δ∂µ φ contribute to δS. ˙ (6. R (6.13) (6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x).t2 ) d3 x . Variations in the action can come from the coordinates or the ﬁelds. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . Consider a lagrangian density L which depends on these functions. dσµ . ˙ x The second term in Eq. which in classical mechanics vanishes at the boundary) does not play a role. . . . i.2. t1 (6. their derivatives and possibly on the position. (6.

21) which diﬀer only by surface terms.2 Lagrangians for spin 0. (6. ∂ (6. 2 (6.25) (6.23) (6.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ. 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).26) The Dirac ﬁeld The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ. (6. (6. the second term is irrelevant. The integrand of the ﬁrst term must vanish.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. 2 ψ = 0.20) (6. δL δL ∂µ = . 2 ∗ (6. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. Using the variations in ψ (in the symmetric form).19) δ(∂µ φ) δφ 6. (2 + M )φ (x) = 0.29) . One easily obtains (2 + M 2 )φ(x) = 0. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found. leading to (2 + M 2 )φ(x) = 0.27) (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. leading to the Euler-Lagrange equations.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration).24) (6.

34) (6.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . The Majorana case is in fact more general.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η). . the mass term in the Dirac lagrangian 6.31 is given by LM (Dirac) = −M ξ † η + η † ξ . 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 . In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). ψL ≡ (ψL )c = C ψL = 0 (6.g. which comes because we in essence use ’real’ spinors. We note that of which the left part coincides . The reasons for Grassmann variables will become clear in the next chapter.31) showing e. 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.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . 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. and thus (β ∗ α)∗ = α∗ β = −βα∗ .7).32) There exists another possibility to write down a mass term with only one kind of ﬁelds. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . (6.38) (6. 2 αβ = −βα. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . namely2 LM (Majorana) = + (6. 2 (6.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. ∂ ∂ 2 2 (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 .g. or equivalently separate the ﬁeld into right. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . but note the factor 1/2 as compared to the Dirac lagrangian. (6. Using the twospinors ξ and η. Peshkin and Schroeder or Exercise 12.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.36) (6. that the lagrangian separates into two independent parts for M = 0.and lefthanded ones. (6.

40) 6.47) .+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . one can consider the variations at the initial or ﬁnal surface.σ1 with σ = (R3 .Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ . (6..42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν .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 . these conserved quantities become operators.18. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0.43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). As an example for the classical ﬁeld. discussed in the next chapter. t). In particular when the lagrangian is invariant under symmetries. we consider the second term in Eq. In this case. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). which is the space-integral over the zero-component of a conserved current. (6.44) δL ∆φ + Θµν (x)δxν . Q(t1 ) = Q(t2 ). δ(∂µ φ) (6. δ(∂µ φ) (6. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. Q= d3 x J 0 (x. t). consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. . ∂µ J µ (x) = 0. + ∂R dσµ dσµ ∂R ≡ where d x . (6. . requires the presence of a conserved quantity.3 Symmetries and conserved (Noether) currents Next.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 .41) (6.. δ(∂µ φ) For continuous transformations. the presence of a symmetry that leaves the lagrangian invariant. Returning to δS the surface term is rewritten to δS = R d4 x . These do not aﬀect the dynamics and will give rise to conserved currents. (6. which in a consistent picture precisely generate the symmetries. (6. one immediately obtains δL = −∂ µ V ν + ∂ ν V µ − λ g µν (∂ρ V ρ ) = −F µν − λ g µν (∂ρ V ρ ) δ(∂µ Vν ) leading to the equations of motion ∂µ F µν + M 2 V ν + λ ∂ ν (∂ · V ) = 2 + M 2 V ν − (1 − λ)∂ ν (∂ · V ) = 0. 6. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) .

3) j µ = i e φ∗ ∂µ φ.g. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . t) is the conserved charge (Q = 0). including e the space-time translations and the Lorentz transformations.51) (6.50) (6. e.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.54) (6. omitting the parameter Λ jµ = − δL δL (i e ψi ) (−i e ψi ) − δ(∂µ ψi ) δ(∂µ ψ i ) 49 = e ψγ µ ψ. = 0. From Noether’s theorem one sees that the current Θµν ǫν is conserved. 2 (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). (6. The integral over the zero-component. H = d3 x Θ00 (x).4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. (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. (6. Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . In the next section we will see the quantity show up as the charge operator.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .49) These currents are conserved as discussed already in chapter 2.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters. ˙ Q = d3 x j 0 (x.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. . 1 (6. (6. we obtain (since δxµ = 0). Summarizing. 6. ↔ (6. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x).53) These are the energy and momentum.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space).

e. however.67) i. 6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ .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 . In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. where H ρµν is the quantity appearing in M ρµν . one has a tensor that satisﬁes (exercise 6. (6. ∂µ T µν (6. −i eΛ(x) ∗ i eΛ(x) (6. the angle of the transformation depends on the space-time point x.61) A ﬁnal note concernes the symmetry properties of Θµν .3). φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. ∂µ φ(x) + i e ∂µ Λ(x) e (6.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. Deﬁning T µν = Θµν − ∂ρ Gρµν .69) i eΛ(x) φ (x). Any lagrangian containing derivatives.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. (6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .59) (6. (6. φ(x). In general this tensor is not symmetric.64) (6. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4.63) µν = ∂µ Θ .g. is not invariant under local gauge transformations. e.3 we have seen global transformations or gauge transformations of the ﬁrst kind.66) In section 6. 2 (6. (6. independent of x.70) . • Vector ﬁeld Aµ : −i Sρσ = aρσ . (6.68) (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).58) (6.6) T µν = T νµ . in which the U (1) transformation involves an angle e Λ.65) M µρσ = T µρ xσ − T µσ xρ .

D ∂ A (6. (6.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.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). For this purpose it is necessary to introduce a vector ﬁeld Aµ .73) (6. −e j µ Aµ = −e ρφ + e j · A. Dµ φ) − Fµν F µν . (6. ∂ 2 4 (6.72) (6. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector ﬁelds to the original lagrangian therefore produces a gauge invariant lagrangian.78) . 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. 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 . As an example consider the Dirac lagrangian. 1 L (φ. (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). ∂µ φ) =⇒ L (φ. 4 (6. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.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 . i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.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 . 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). (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.

(c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). 2 2 (b) Show that the Klein-Gordon equation for the complex scalar ﬁeld (considering φ and φ∗ as independent ﬁelds can be derived from the lagrangian density L = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ. and deduce what is the charge of the antiparticle as compared to a particle.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 . where j µ = e ψγ µ ψ. φ → eiΛ φ.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.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . i. ∂µ F µν = j ν . ↔ Exercise 6. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0.79) Exercises Exercise 6. give the equation which is satisﬁed by ψ c = Cψ . giving the Maxwell equation coupling to the electromagnetic current.e. ∂µ F µν = j ν . (6. Exercise 6. −eψγ ν ψ.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds. . Exercise 6.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .

e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r.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. Comment: this relation is known as the Pauli equation. .Classical lagrangian ﬁeld theory 53 Exercise 6.4. in order to show that Θρσ − Θσρ = ∂µ H µρσ . 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. use ∂µ T µν = 0 and ∂µ M µρσ = 0. A) (see also exercise 6.6 (optional) Prove the properties of the tensor T µν in section 6. t) = Enr ψu (r. 2me Exercise 6. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . t). 2M 2M with Enr ≡ E − M ≪ M .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. This relation is known as the Pauli equation.

there are conserved ˙ ˙ quantities Q (Eq. q) = ˙ 1 mq 2 − V (q). B]P = dB dA dA dB − . This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. ˙ L(q.11) found through Noether’s theorem. is the Poisson bracket [A. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. q) = p q − L(q. t). in our example p = mq. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. (7.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t). 6. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. p]P = 1. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. H]P dt and dO = [O.Chapter 7 Quantization of ﬁelds 7. (7.4) 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 .3) For variations δO(p. starting from the lagrangian L(q. [q. (7. ˙ 2 (7. dq The basic Poisson bracket is actually the one between q and its canonical momentum p.5) ∆ x ∆3 x 54 . dλ Examples are dO = [O. p]P . 1 qx (t) = 3 d3 x φ(x. An immediate generalization for ﬁelds can be obtained by considering them as coordinates.2) H(p. The bilinear product for the conserved quantities. labeled by the position. q(q. Further. dq dp dq dp (7. however. q) also considered in the previous chapter. namely [q.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). p] = i.

• Poincar´ invariance e The full variation of the ﬁelds. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . t)] = i δ 3 (x − y). φ(y.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . a) = Rj (Λ−1 ) φj (Λ−1 x − a). formally. t). a) φi (x)U (Λ.7) d3 x L (φ(x). Π(y. i U † (Λ. φ. Sometimes it is useful to keep in mind that. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). qx+∆x ) .Quantization of ﬁelds 55 etc. t)] = [Π(x. Π(y. δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 .16) . ∇φ) (7. the discretization procedure above is an explicit example. As an example.11) (7. ∆φ.14) In fact. t). (7. we have L (x) = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 2 (∂0 φ) − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ. qx . Φ(f ) = d4 x φ(x) f (x).2. (7. albeit with ’sharp’ functions. The essential conditions to consistently quantize a theory then are the following.12) (7. discussed in the section 6. (7.15) with furthermore the relations [φ(x. ∂µ φ(x)) = ˙ d3 x L (φ. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta. for the scalar ﬁeld theory. ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ).8) (7. ˙ δ φ(x) (7. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x. 2 2 2 (7. t)] = 0. t).13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory.6) ∆3 x L x (qx . ∂ qx ˙ ∂ qx ˙ (7. for symmetry transformations.10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.

i(xµ ∂ν − xν ∂µ )φ (x) + i (7.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. Φ(g)] = 0.Quantization of ﬁelds 56 or if U (Λ. 2 i ωµν (S µν )i . Mµν ] = i i∂µ φi (x). 2 2 (7. a] = [a† . and [N. P ] = i. [N. (7.18) with S µν given in 6. sometimes referred to as second quantization. [φ(x). • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). [a.23) The hamiltonian in this case can be expressed in the number operator N = a† a. a† ] = 0. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . 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.24) It is straightforward to ﬁnd the commutation relations between N and a and a† . In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . P ] 2 (7. Q] = [P.25) . Pµ ] = [φ (x). 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. a† ] = 1. The measurements cannot inﬂuence each other or [Φ(f ). j (7.17) (Sµν )i φj (x). [Q. (7. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. φ(y)] = 0 if (x − y)2 < 0. This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the ﬁelds through Noether’s theorem.4. Microscopic causality implies local commutativity. ω a† a + 2 2 ω i − [Q. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν .19) 7. a] = −a. P ] = 0 (7. (7.

28) (7.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. [φ(x). a(k ′ )] = [a† (k). (7. is equivalent with [a(k). φ(x )]x0 =x′0 = [Π(x). 7. i. † √ a and a act as raising and lowering operators.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).26) a(k) e−i k·x + a† (k) ei k·x . Π(x )]x0 =x′0 = 0. In this way one has found for the harmonic oscillator the spectrum of eigenstates |n (with n a non-negative integer) with En = (n + 1/2)ω.e.30) [φ(x). (n − 1) a|n † i.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . the real scalar ﬁeld φ(x) becomes d3 k (7. and we see that a state |0 must exist for which N |0 = a|0 = 0. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).g. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them.like’ commutation relations between a(k) and a† (k ′ ).33) . From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . e. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. a† (k ′ )] = 0. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n . but one needs to realize that in that case k 0 = Ek = k2 + M 2 .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. which represents the ’number of particles’ with momentum k.32) (7.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. and (7. Note that we will often write a(k) or a† (k). 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 . (2π)3 2Ek (7. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.e.

35) where the vacuum contribution disappears because of rotational symmetry. . (2π)3 2Ek (7.45) (7. For the momentum operator one has Pi = d3 x Θ0i (x) = 1 2 d3 x ∂ {0 φ ∂ i} φ = d3 k k i N (k). .37) (7. n1 ! n2 ! (7. |0 . i. is solved by subtracting it as an (inﬁnite) constant. (2π)3 (7.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). writing all annihilation operators to the right of the creation operators. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k).40) This procedure is known as normal ordering. (2π)3 2Ek (7.34) where V = (2π)3 δ 3 (0) is the space-volume. . and multiparticle states |(k1 )n1 (k2 )n2 . For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. a(k)|0 = 0.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. (2π)3 2Ek (7. which now contains an inﬁnite number of oscillators. (2π)3 2Ek d3 k a† (k) ei k·x . 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.e.41) (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). d3 k |k k|. (2π)3 2Ek (7.42) (7.46) (7. assuring that the vacuum (by deﬁnition) has eigenvalue 0!.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. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). . . deﬁning particle states |k = |k (with positive energy. d3 k a(k) e−i k·x . which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ .44) p|φ− (x)|0 = ei p·x . k 0 = Ek = k2 + M 2 ). |k = a† (k)|0 .39) The problem with the vacuum or zero-point energy. The rest of the states are then obtained from the groundstate via the creation operator. This term will be adressed below.

53) t=0 .49) (7. i.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). that they satisfy the required commutation relations in Eqs (7. d3 k = (2π)3 2Ek = (7. d3 k ei k·(x−y) = −i∆+ (y − x).e.51) (7. = − (2π)3 2Ek (7.48) (7. (iii) ∆(0. (2π)4 (7. φ+ (y)] d3 k e−i k·(x−y) . (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. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (7.56) = −δ 3 (x).18).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.17) and (7. φ− (y)] ≡ i∆+ (x − y). ∆(x) = − ǫ(x0 ) δ(x2 ). φ(y)] consider µν [φ+ (x). (7. (2π)3 2Ek ≡ i∆− (x − y). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. x) = 0 and hence ∆(x) = 0 for x2 < 0.52) The result for the invariant commutator function is [φ(x).50) or as integrals over d4 k. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). The last item to be checked for the scalar ﬁeld are the causality condition.47) ′ d3 k ′ e−i k·x+i k ·y [a(k). In order to calculate [φ(x).

66) . a† (k ′ )] = [b(k). 2 From the lagrangian density (7. extending it with the charge operator and the introduction of particle and antiparticle operators. ↔ (7.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). which is equivalent to the relations (only nonzero ones) [a(k).61) (7.57) (7.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.e. 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) .59) (7.60) = i φ∗ ∂µ φ. we will consider it as a repetition of the quantization procedure. φ† (y)] = i∆(x − y).62) (7.65) 7.63) i. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). = ∂{µ φ∗ ∂ν} φ − L gµν .Quantization of ﬁelds 60 7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (2π)3 2Ek (7. (2π)3 2Ek (7. The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) . b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). (2π)3 2Ek (7. (7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.

s). s) ei k·(x−y) x0 =y 0 . s)u† (k. s)e−i k·x + d† (k. d† (k ′ .78) (7. The quantized ﬁelds are written ψ(x) = s d3 k b(k. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .69) (7.67) i ψ / ψ − M ψψ gµν . (7. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. ↔ i ψγµ ∂ν ψ − 2 (7. † {ψi (x).68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). s) e−i k·(x−y) j s † + vi (k. i.73) (7. (2π)3 2Ek (7. jµ Θµν = = ψγµ ψ. s)¯(k.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : .71) (7. . u v (2π)3 2Ek (7. (7. b† (k ′ . The solution is the introduction of anticommutation relations.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. s)u(k. s)vj (k.75) (7.e.74) (7. then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . s)e−i k·x . (2π)3 2Ek d3 k b† (k. ∂ 2 ↔ (7. s′ )} = {d(k.70) where the last line is obtained by using the Dirac equation.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. s)¯(k. (2π)3 2Ek (7.77) Note that achieving normal ordering. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k. s)ei k·x . s)ei k·x + d(k. interchanging creation and annihilation operators. s). s)v(k. {b(k.

ψ j (y)} at arbitrary times one has {ψi (x). 4 Π0 Πi δL = 0. ∂ (7. 7.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before.86) (7. but has the problem of being noncovariant. ˙ δ A0 δL = F i0 = E i . ψ j (y)] = (i/x + M )ij i∆1 (x − y). we would have obtained [ψi (x). which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem).83) From the lagrangian density the canonical momenta are = = (7. 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. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.80) 3 = δ 3 (x − y) δij . but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. −λ(∂ρ Aρ ).85) which reﬂects the gauge freedom. (7. ∂ (7.82) where i∆1 = i∆+ −i∆− .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.88) . It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . (7. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. 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). For the scalar combination {ψi (x). 4 2 This gives the equations of motion discussed before. ˙ δ Ai (7.6 The electromagnetic ﬁeld 1 L = − Fµν F µν . one has † {ψi (x).84) (7. as the vanishing of Π0 induces a constraint.

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. containing a time-like photon. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. It gives 3 k µ ǫ(λ) (k)c(k. µ λ=0 (7. rj ]P .90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). Aν (y)] = [Πµ (x). 2. 0. i. and are equivalent with [c(k. Πν (y)] = 0. The canonical equal time commutation relations are [Aµ (x). λ) − c† (k. a longitudinal photon and two transverse photons.11. ′ (7. 7. (b) Proof for a quantity O(p. pj ]P . one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.94) Choosing k µ = (|k 3 |. These problems are solved by the Lorentz constraint between physical states given above.e. q) that dO/dλ = [O. λ)c(k. what are [ri . 0)c(k. and [ℓi . (7. Q]P . c† (k ′ . where Q is the conserved quantity in Eq. µ λ=0 (7. (7. pj ]P .1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq.3) satisfy the conditions of a Lie algebra. λ)e−i k·x + c† (k. for which it is suﬃcient that ∂µ Aµ |A = 0. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. Aν (y)]x0 =y0 = i gµν δ 3 (x − y). k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. 0.92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i. j = 1. i.93) which does exhibit problems with the time-like photon. (d) Use Poisson brackets to calculate q and p.95) Exercises Exercise 7. 0) . [ℓi .Quantization of ﬁelds 63 for physical states |A and |B .91) (7. λ)ei k·x . λ). λ)|A = 0. Πν (y)]x0 =y0 = i gµν δ 3 (x − y).89) with four independent vectors ǫµ . In fact the commutation relations imply ˙ [Aµ (x). which reproduce the Hamilton equations. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. ˙ ˙ . (7. 6. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ).e.

Pµ ] = i∂µ φ(x). (e) Show that ∆R (x) = 0 if x0 < 0. (b) Check the above commutation relation [φ(x). Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. It is invariant. ∆hom (Λx) = ∆hom (x) if f is an invariant function. (c) What are the poles in the integral under (b) in the (complex) k0 plane? (d) Depending on the paths in the k0 plane going from k0 = −∞ to k0 = +∞ one can distinguish four diﬀerent Green’s functions (solutions to the above inhomogeneous Klein-Gordon equation): k0 -E E ∆ R (retarded) ∆ C (causal) ∆ ∆A (advanced) Show that by appropriate shifts of the poles into the complex plane. (2 + M 2 ) ∆hom (x) = 0.2 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. we can write for ∆R an integral where the k 0 integration remains along the real axis. e−i a requires [φ(x). (2 + M 2 ) ∆inhom (x) = −δ 4 (x).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. ∆R = 1 (2π)4 d4 k e−ikx k2 1 − M 2 + iǫk 0 (with the implicit prescription to take ǫ → 0 after integration). (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. f (Λk) = f (k) for Lorentz transformations Λ. . Give also the expression for ∆C . (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). Pµ ] for the (real) scalar ﬁeld using the expressions for the ﬁelds and momentum operator in terms of creation and annihilation operators.Quantization of ﬁelds 64 Exercise 7.3 (a) Show that the correct implimentation of symmetries in the Hilbert space.

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

1 Parity xµ = (t. T. (iγ µ ∂µ − M ) ψ(x) = 0. starting with the Dirac equation for ψ(x). Table 8. 8. time reversal (T) and charge conjugation (C). T and C for classical quantities is shown in the table 1. x (8. The consequences of P. ˜ (8.1: The behavior of classical quantities under P.2) where ηP is the intrinsic parity of the ﬁeld and A is a 4 × 4 matrix acting in the spinor space.Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries. Both ψ p and ψ satisfy the Dirac equation. r) −→ xµ ≡ xµ = (t. We can determine A. parity (P). writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). −r).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 .

(2π)3 2Ek (8. −λ)v(k.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. λ) e−ik·x − d† (k. m) γ0 v(k.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. −λ) e−ik·x − d† (k. (i˜ µ ∂µ − M ) ψ(˜) = 0.10) i. iγ µ† ∂µ − M ψ(˜) = 0.5) (8. one can also consider the scalar ﬁeld and vector ﬁelds.3) (8.e. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. −λ) e−ik·x − d† (k. −λ). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. λ)γ0 v(k. x (8. λ)γ0 u(k.7) = λ d k x x ηP b(k. γ x (8. λ) e−ik·˜ + d† (k. λ)v(k. . −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. λ) Pop = ηP b(k. λ) eik·x . −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜).6) (check this for the standard representation. x The latter behavior of the vector ﬁeld will be discussed further below. −λ). m).9) (8.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. the above operation reverses the sign of λ). if helicity λ is used instead of the z-component of the spin m. ˜ Pop d(k. λ) P −1 = −η ∗ d(k. λ)u(k.11) (8. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ)u(k. m). 8. −λ)u(k. ˜ −v(k. m) = = ˜ u(k. op P (8. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). λ)v(k.4) (8. In the same way as the Fermion ﬁeld.

antilinear operator is anti-unitary if A† = A−1 . u(k. r) −→ −˜µ ≡ −xµ = (−t. λ) Cop = ηC d(k. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . x (8. λ)u(k. λ) eik·x . i. λ) −1 Cop = ∗ ηC b(k. m). Top |φ = φt | = (|φt )∗ . The same relations hold for helicity states. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0.e.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). (8. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . to ﬁnd ψ t (−˜) that is a solution of the Dirac equation. x (8. . λ)v(k. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. m). x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0.19) Cop d(k.21) where A is a 4 × 4 matrix acting in the spinor space.18) (8. As time reversal will transform ’bra’ into ’ket’. m) ¯ C v T (k.3 Time reversal xµ = (t. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0.16) d k ηC d(k. λ)C uT (k. λ)C v T (k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8.Discrete symmetries 68 the latter being also a solution of the Dirac equation. The (time-reversed) Dirac equation becomes.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . λ) e−ik·x + b† (k. (2π)3 2Ek (8. we start with the complex conjugated x Dirac equation for ψ. r). x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. (8.14) (8. 8. m) ¯ = = v(k. as discussed in section 4). x x (8. λ). γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0. Norm conservation requires Top to be anti-unitary2 .17) 3 = λ This shows that −1 Cop b(k. λ) e−ik·x + b† (k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k.15) (where one must be aware of the choice of spinors made in the expansion. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). λ). it is antilinear1 . x iγ µT ∂µ − M ψ(−˜) = 0.

λ| a. 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) (8.24) (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(−˜). (8. λ) T −1 = η ∗ d(k. λ) eik·x + d† (k. −p. ∗ ˜ v (k. there are 16 independent of these bilinear combinations. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate.26) d k x x ηT b(k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ) i γ5 C v(k.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k. λ) eik·x + d† (k. λ)u∗ (k.32) (8. λ)iγ5 Cu(k. a state |a. −p. λ)u∗ (k. ˜ = 0|b(k)|A 8. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. −p.28) (8.29) In table 2 the behavior of particle states under the various transformations has been summarized. p. ˜ Top d(k. λ| ¯ C |¯. λ). λ)iγ5 Cv(k. x (8.25) (check this for the standard representation). λ) e−ik·x (2π)3 2Ek (8.34) . p. λ)u∗ (k. λ)v ∗ (k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. p. Since there are 16 independent 4 × 4 matrices. −λ a T a.31) (8. 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) Top = ηT b(k. −p.33) (8.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k.Discrete symmetries 69 Table 8. The result is = = λ (8. λ). λ a |a.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. op T (8.30) (8. λ a P |a. λ |¯. λ)v ∗ (k. λ) eik·x + d† (k. λ). λ) eik·˜ + d† (k. λ)v ∗ (k. λ). −λ |¯. p.

¯ ′ ′ (8.40) where the coeﬃcients are functions of invariants constructed out of the momenta. In many applications the expectation values of currents are needed.. V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . (q 2 ). momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ).38) (8. 8.35) The x-dependence can be accounted for straightforwardly using translation invariance. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. (8. There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ).Discrete symmetries 70 Table 8.37) (8. 2M (8. (8. p′ |V µ (x)|p = u(p′ )Γµ (p′ . p) can be built from any combination of Dirac matrices (1. as well as under the combined operation Θ = PCT (see Table 3). p′ . γ µ . F. s . of which the expectation value between momentum eigenstates can be simply found. (q 2 ).36) (8. Note that the lagrangian density L (x) → L (−x) under Θ. C.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). for the vector current V µ (x). and T. e. called form factors. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).5 Form factors Currents play an important role in ﬁeld theory.. 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 . F. γ ν ].41) 2M 2M . As an example consider the vector current for a point fermion. in this case only q 2 = (p′ − p)2 . s′ |V µ (x)|p. For instance for nucleons one has Γµ (p′ . p)u(p) e−i (p−p )·x . ¯ V µ (x) =: ψ(x)γ µ ψ(x) :.g. ¯ In general the expectation value between momentum states can be more complicated.39) where Γµ (p′ . σ µν . The 16 combinations of Dirac matrices appearing above are linearly independent. p′ |V µ (0)|p = u(p′ )γ µ u(p). C. γ5 γ µ or γ5 ).

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

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

QH (t)|q. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t ≡ q|q.2) (ii) Heisenberg picture. t0 ).3) (9. in which the operators are time-independent. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0.6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. |ψS (t) = U (t. Consider the two (time-independent) Heisenberg states: |q. AS (t) = AS and the states o are time dependent. H]. t0 ) AH (t0 ) U (t. [AH . (9.5) (9. (9. t ′ QH (t′ )|q ′ . in which the states are time-independent. t . |ψH (t) = |ψH . t0 ) = H U (t. t′ ≡ q ′ |q ′ . t ′ |q . t′ . and the operators are time-dependent. AH (t) = U −1 (t.4) ∂ U (t.7) 73 . (9.Chapter 9 Path integrals and quantum mechanics 9. t0 ) ∂t (9. ≡ 0. t0 )|ψS (t0 ) . i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . (9.1 Time evolution as path integral U (t.

B]] + [[A. qj ). n |q (9. tj+1 |qj . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . 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|. . B] + [A. t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . |q1 q1 |e−i H∆τ |q .12) where the transformation between coordinate and momentum space involves q|p = ei p.11) (9. Then qj+1 . B] + . t′ |q.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q .qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . 2π (9. (9. . The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. t = q ′ |e−i H(t −t) |q . [A. tj+1 |qj . QS |q ′ ≡ q ′ |q ′ . eA eB = eC with C = A+B+ (9.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . . qj )]) . ˙ 2π 1 1 1 [A. which is a hamiltonian function in which the operators are replaced by real-numbered variables..q .8) Choose t as the starting point with |q = |q. H(P. (9.13) At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential.14) . 2 12 12 (9. ′ (9. . B].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 ′ . Combining the two terms gives qj+1 . Q) = K(P ) + V (Q) = (P 2 /2M ) + V (Q) one can split e−i H∆τ ≈ e−i K∆τ e−i V ∆τ .. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . 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 .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 . Q) expressed in terms of the operators P and Q.16) this. The hamiltonian is an operator H = H(P. 2π (9. use the Campbell-Baker-Hausdorﬀ formula. with the correction1 being of order (∆τ )2 . t′ |q.

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

(9.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.24) (9.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . 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.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function.21) again a multidimensional integral. α(x) = n αn fn and K(x.27) (9. (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K.25) (see Exercises).Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. N (9. . with Dα F [α] = n dαn N F (αn ) (9. Dα F [α] = Dα F [α + β]. 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. A useful property of functional integration is the translation invariance. Note that procedure (i) is an example of the more general procedure under (ii).n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.23) Having deﬁned the Gaussian integral. 2 (9. Consider the gaussian functional as an example. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. y) = ∗ m. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m.

In principle the deﬁnitions of functionals is the same.g. δα(x) (9. i.e. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. α(y) = δ(x − y). αy = δxy . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . which vanish. θ2 = 0. y) θ(y) . δα(x) δα(x) δ αKβ = dy K(x. F [θ. θη = −ηθ.32) (9. δα(x) or in discretized form δ 1 ∂ ∂ with .29) (9.30) (9. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K.33) (9.37) 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. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . Examples of functional diﬀerentiation are (9. products of the same pair appear.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. e.38) .35) For applications to fermion ﬁelds.31) δ δ α = dy α(y) = dy δ(x − y) = 1. −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). αn = δmn . (9. n θn fn (x) dx θ∗ (x) θ(x) . In the expansion of the exponential (from second to third line). (9. dx dy θ∗ (x) K(x. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. however.34) (9. the Gaussian-type functionals.36) where the coeﬃ- This gives for the Gaussian functional integral after expanding θ(x) = cients θn are Grassmann variables the result Dθ∗ Dθ exp (−θ∗ θ) = n ∗ dθn dθn ∗ dθn n exp − ∗ θn θn n = = n ∗ dθn exp (−θn θn ) ∗ ∗ dθn dθn (1 − θn θn ) = 1. (9.

Path integrals and quantum mechanics

78

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

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

= δ(x − y),

(9.39)

(9.40) (9.41)

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

9.3

Time ordered products of operators and path integrals

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

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

q ′ , t′ |q, t =

Dq exp i

t

dτ L(q, q) , ˙

(9.44)

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

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

t′

Dq exp i

t i 2

dτ Leﬀ (q, q) , ˙

(9.45)

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

i

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

δ(0) ln f (q).

dqi q ′ , t′ |qn , tn . . . (9.46)

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

t′ t′ t

dτ [pq − H(p, q)] ˙

(9.47) (9.48)

Dq q(s) exp i

t

dτ L(q, q) . ˙

Path integrals and quantum mechanics

79

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

t t′

(9.49)

dτ L(q, q) . ˙

(9.50)

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

9.4

An application: time-dependent perturbation theory

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

′

(9.51) (9.52) (9.53)

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

t′ ∞ n=0 t′

= HI (t) ψI (t), = [AI , H0 ],

(9.54) (9.55)

(9.56) (9.57)

dτ HI (τ ) U (τ, t),

t

(9.58)

U (n) (t′ , t),

(9.59)

1−i (−i)

t t′ t

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

**Path integrals and quantum mechanics
**

t′ t t′ t

80

U

(2)

(t , t) = 1 − i = (−i) . . .

′

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

t′

U

(n)

(t , t) = (−i)

t

′

dτ HI (τ ) U (n−1) (τ, t)

t′ τ1 τn−1

= (−i)n

t

dτ1

t t′ t′

dτ2 . . .

t t′

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

t

=

(−i)n n!

dτ1

t t

dτ2 . . .

(9.60)

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

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

t′ τ1

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )

t′ t′

+

1 2

dτ2

t τ2

dτ1 HI (τ1 ) HI (τ2 ),

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

t′ t′

dτ1

t t

dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 )

t′

+

1 2

t′

dτ2

t t

dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ),

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

t′ t′

dτ1

t t

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

the desired result.

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

∞ t′ t′ t′

dτ1

t t′ t

dτ2 . . .

t

dτn T HI (τ1 ) . . . HI (τn ) (9.61)

≡ T exp −i

dτ HI (τ ) .

t

t′ |q. t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. t = = ′ ′ V = q . i. q) → L(q. t = ei Ht |q . say.63) for plane waves. i.s. T ′ |q ′ . φn (q. t q. the creation of an electron (think of a radio-tube. ˙ ˙ physically pictured as. L(q. for g.66) . t) φ∗ (Q) e+i En T .62) p Dq D exp i 2π t dτ [pq − H(p. Considering the source to be present between times t and t′ . which in turn are embedded between an early time T and a future time T ′ . n (9. T ′ |Q. t|n = q|n e−iEn t with q|n the time-independent wave function. Before and after these processes there is only the vacuum or ground-state |0 .e. T = n J t′ = = Dq exp i t dτ [L(q. We can again rewrite the time-ordered products as functional integrals as derived in the previous section.64) (9. q) + J(τ )q] . q)] ˙ (9. 9. q) + J(t) · q. furthermore. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). q)] . harmonic oscillator. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. The Heisenberg state |q. making the electron) and the absorption of an electron (think of a detector). t|Q. T < t < t′ < T ′ . it is possible to switch on an interaction.e. a set |n of physical eigenstates. t |q. t|n n|Q. Consider.Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9.65) J dq ′ dq Q′ .1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. t′ |q. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. for the physical states o q.5 The generating functional for time ordered products By introducing a source-term. t′ q ′ . t Q′ . one has q ′ . q . t is related to the Schr¨dinger state |q by |q. t|Q. T = n q. ˙ J (9. T . T with q.

(9.67) = = φ0 (q. t′ |q. J (9. Since we have (neglecting multiplicative factors). t) (9.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. T ′ |Q. thus T →i∞ lim q.68) (9.6 Euclidean formulation The above expression (Eq. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. dq dτ + J(τ )q(τ ) . = (i)n 0|T Q(t1 ) .70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . t′ ) q ′ . The importance of Z[J] is that the time ordered product of operators can be obtained from it. . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. J=0 (9. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q. δJ(tn ) hence the name generating functional. eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . .69) The factor that has been divided out in the ﬁrst line of this equation is a numerical factor depending on the boundaries of the space-time volume (T and T ′ ). . Better is the use ˜ of imaginary time t = −it.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. t). . t 0 0out |0in J . T = φ0 (q. T → i ∞ and T ′ → −i ∞. 9. ˜ t′ → ∞. . 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.70) for the generating functional is in fact ill-deﬁned. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ .71) 9. Q(tn )|0 .

dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). for Grassmann valued functions.and Grassmann-valued functions. ˜ ˜ ZE [J] δJ(t1 ) .Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. . i dq . Exercises Exercise 9. δJ(tn ) J = 0 ˜ t = it (9. when L q.75) = − LE q. ˜ dt (9. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. (9.77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors.n fm (x) Kmn fn (y). . .72) (9. The diﬀerentiations in Z[J] and ZE [J] are related. ensuring convergence for the functional integral ZE [J]. .2 Check the examples of functional derivation for real-. e. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q.g. y) = ∗ m. . i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . where Kp are the eigenvalues of the matrix Kmn in K(x. det K Exercise 9.73) d˜ LE q. 1 = √ .74) − V (q) 2 (9. complex. ZE [J] = = where LE q. (9. As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . + V (q). δ n Z[J] 1 Z[J] δJ(t1 ) .76) which is a positive deﬁnite quantity. δJ(tn ) = (i)n J=0 1 δ n ZE [J] . τ dq ˜ dt ≡ −L q. det K 1 .1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions.

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

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

which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . d4 xn G(n) (x1 .9) i.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x .10) J=0 In order to determine ∆F . xn ) J(x1 ) . . .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . . one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. (10. 2 2 → (10.e. (2π)4 (k 0 + iǫ)2 − E 2 (10. These Green functions appear in the expansion Z[J] = n in n! d4 x1 . . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. . .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). (10. As discussed in the previous section (exercise 9.5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 .13) .3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . . Furthermore. J(xn ). consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). (10.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. . xn ) = x2 xn . with Z0 [0] = 1.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. .

16) C where C is a closed contour in the complex k 0 -plane. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . φ+ (0)] = (2π)3 2E are also shown in exercise (7. t) = 0 for t > 0.e. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (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 . The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . 10.4).23) k0 =i∞ k0 =−i∞ d4 k e−i k·x .21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (10. (2π)4 k 2 − M 2 (10.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. 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). It is straightforward (closing contours in the appropriate half of the complex plane) to prove for instance that i ∆C (x) = θ(x0 ) i ∆+ (x) − θ(−x0 ) i ∆− (x). The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). (2π)4 k 2 − M 2 (10. . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. 4 k 2 − M 2 + iǫ (2π) (10.14) They are solutions of the inhomogeneous equation. t) = 0 for t < 0. the advanced Green’s function ∆A (x) = satisfying ∆A (x. i. The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. 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.20) = exp − 2 where or (see ﬁg.22) (10. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).15) e−i k·x d4 k .17) d4 k e−i k·x .18) Thus we see various solutions. ∆F = ∆C . (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . (2π)4 (k 0 − iǫ)2 − E 2 (10. ∆(x) = − d4 k e−i k·x . (10. µ 1 1 • Firstly.

x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . = θ(x0 − y 0 ) 0|[φ+ (x). φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.25) x1 x2 x3 = x4 + + . we have G0 (x1 . . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. We ﬁnd (see also section 7. . 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.24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. x3 . 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. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). or diagrammatically (4) (10. the same result as above. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] .3) i ∆F (x − y) = 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0 = i ∆C (x − y). x2 .1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility.

27) When interactions are present. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].28) 1 exp − i 2 = exp 1 2 d4 x d4 y exp i δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)] φ=0 .Feynman diagrams for scattering amplitudes 89 10.32) (10.33) (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 .g ) (x1 . x2 .g ) (x1 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10.2 Generating functionals for interacting scalar ﬁelds d4 x [L0 (φ) + LI (φ) + Jφ] 1 δ i δJ(z) d4 x [L0 (φ) + Jφ] (10. ≡ −i g d4 z (10.26) 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. Z[J] = = Dφ exp i exp i d4 z LI (10. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J .29) This expression will be the one from which Feynman rules will be derived.e.30) and in zeroth order G(0. x2 ) = i ∆F (x1 − x2 ) = G(4. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. x3 . i. (10. (10.g 0 ) = G0 = 1. x2 . (2) (0) (10. x3 . (10. the generating functional can be written. L (φ) = L0 (φ) + LI (φ).35) z .34) Introducing a vertex point to which four propagators are connected.31) G(2. x2 ) = G0 (x1 . x4 ) = G0 (x1 .

g ) (x1 . . . . i.39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. .40) in−1 n! d4 x1 . . J(xn ). Diagrammatically this exponent only contained the two-point function. . . In the free case we have seen that the exponent of Z[J] contained all essential information. (10. . . δJ(xn ) . .Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. This remains also true for the interacting case. did not contain parts that could be written as products of simpler Green’s functions. Z[J] = n in n! d4 x1 . G(n) (x1 . J(xn ) c (10. .41) in terms of socalled connected Green’s functions. . (10. xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . (10. which also was the only Green’s function for which the diagram was connected. .36) J J from which one obtains G(0. . . . (10. To see this write Z[J] = exp (i W [J]) with (by deﬁnition) the expansion W [J] = n or i W [J] = ln Z[J]. . . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . J=0 (10. .g 1 ) = 1 8 z 1 8 .37) G(2. .38) The result for the 4-points Green’s function is left as an exercise. xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (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 ) J(x1 ) . .e. . xn ) J(x1 ) . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . d4 xn G(n) (x1 . . d4 xn G(n) (x1 . . (10.42) x1 G(n) (x1 .

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

in = β. in ⇔ β. out . then (n) (n) Gsc = Gc for n ≤ 3. 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 . x1 G(4. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout .e. in = |p1 . out|p′ . out . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y).46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. in| (choose ϕ0 = 0).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 ). (2) (10. x2 . x3 . in|p′ . . ∂µ ∂ µ + M 2 φin (x) = 0. in|SS † |α. (10.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. . in|S|p′ . Proof: 0. in = p. out| = eiϕ0 0. out|p′ . . φout and the interpolating ﬁeld φ(x). For the free scalar theory. . out (suppressing except momenta all other 1 m quantum numbers).g ) (x1 . implying the absence of tadpoles. x4 10. Sβα ≡ β. pn . . the vacuum expectation value of the ﬁeld φ(x) is zero. . in|S = 0. (10. in into ﬁnal state particle states (out-states) |β. (10. in = p.g. in = |α. out| and S † |α. p. out|α.47) . thus β.Feynman diagrams for scattering amplitudes 92 i. one ﬁnds for the connected 4-point Green’s function at order g.44) J=0 For the interacting theory. e. translation invariance). this will be translated to the action on ﬁelds. out = δαβ ↔ SS † = 1. out = |p′ . in|S = α. Proof: α. (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). in = p. x4 ) = c 1 x3 z (10. considered explicitly. (2) The one-particle state is invariant (conservation of energy and momentum. Next. out| ⇔ |α. in = S|α. out|α. out = p|p′ . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin .45) x2 as the only surviving diagram (see exercises). p′ . in|S = β.

z) φ(z) d4 y d4 z ∆A (x − y) K(y. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . The convergence therefore must be weakened to t→−∞ √ (10. S is invariant under e Poincar´ transformations: U (Λ. (2π)3 2E (10. Considering plane waves one sees from √ √ 0|φ(x)|p = lim Z 0|φout (x)|p = Z e−i p·x t→∞ √ √ = lim Z 0|φin (x)|p = Z e−i p·x t→−∞ ↔ that identical normalization of (single-particle) plane waves in initial and ﬁnal state implies the same wave function normalization Z for in and out ﬁelds. (10.53) 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. (10. β. The relation between S and Z[J] To establish this relation. a)SU −1 (Λ.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. out|φout = β. The time evolution operator (see Eq. as it stands. in|φin S → φin S = S φout (x).54) . in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. 2 ∂µ ∂ µ + M0 φ(x) = J(x).62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . i.g. √ Although the above.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. z) φ(z). e. implies the strong (operator) convergence φ(x) → Zφin (x). this can actually not be used as it would imply [φ(x). which in a later stage (renormalization) will become more important. φ(y)] = Z [φin (x). 9.51) (10.and out-ﬁelds is: φin (x) = S φout (x)S −1 . The relation between S-matrix and in. a) = S. the source term Jφ in the lagrangian density is treated in the interaction picture.e.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 ). a causality condition that can be proven to imply the absence of interactions.52) where Z is a constant.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. 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. (10. φin (y)] = iZ ∆(x − y). 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.

t i Ut∞ [J] φ(x)U−∞ [J] (10. just as we did for φ(x). δJ(z) (10. (10. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. eB ] = [A. φ− f (10.e. C] are c-numbers. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. φ(x).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).60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). z) . 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.59) (10.61) e φ+ f where the normal ordered expression : e φ f : is used. S U [J]] = √ Z δ S U [J]. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. φ(x). Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. (10. e = i.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. e e ] = [A. express it in terms of advanced and retarded Green’s functions. we can.62) where F [J] is some (arbitrary) functional. B + C]e e . B] (for the case that [A. z) d4 y d4 z ∆(x − y)K(y. B] is a c-number).63) . Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). [A. δJ(z) (10. with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. z) δU [J] . provided [A. δJ(z) Taking the diﬀerence between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. φ(y)] f (y) e e . which is equal to the expression e in which the creation operators are placed left of annihilation operators. : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. y) δ δJ(y) : F [J]. B]eB . B] and [A.58) In order to ﬁnd a solution to this equation note that. one has [A.57) i [φin (x).

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

but overall momentum conservation at a vertex is understood. to be precise iLI vertices in momentum space are introduced. Diagrammar) in the case of the interaction terms g f (10. Note that in calculating amputated Green’s functions external lines are neglected. At these vertices each line can be assigned a momentum. ’t Hooft and M.66 and 10. If problems arise go back to the deﬁning expression for the generating function in 10. the factor 2 corresponds to the two-points Green’s function having two identical ends).27. for example up to order g 2 the scattering of two neutral spin 0 particles (real scalar ﬁeld) is described by + + + (Rule 2) The contribution of each diagram is obtained by multiplying the contributions from propagators. Veltman. vertices and external particle wave functions in that diagram. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. which cancels the factor 1/2 in the quadratic piece. 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 10.69) LI (φ) = − φ3 − φ4 . = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). = −i g . For the interaction terms in the lagrangian. or calculating full Green’s functions external lines are treated as propagators.Feynman diagrams for scattering amplitudes 96 (in fact the inverse of the operator found in the quadratic term is for real scalar ﬁelds also multiplied by a factor 2.27. 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. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines ﬁxed). For the symmetry factor consider the examples (given in G.

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

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

Feynman diagrams for scattering amplitudes 99 The result is which contains an extra minus sign as compared to a bosonic loop. Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge). The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z). λ 1 ∂ (10. writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached. 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 .74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . + 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. i p p i ui (p) ui (p) ¯ vi (p) ¯ incoming fermion outgoing fermion incoming antifermion outgoing antifermion i p p i vi (p) In writing down the expressions for Feynman diagrams one has to be aware that the wave functions and the propagators have more than one component. It is necessary to start at the end of a fermion line (with the above arrow convention an outgoing fermion wave function u or an incoming antifermion ¯ wave function v ) and keep on following that line. 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. Vector ﬁelds and Quantum Electrodynamics As the most important example of vector ﬁelds consider the lagrangian density for quantum electrodynamics (QED). 4 2 In addition to the fermion propagator and fermion wave functions discussed in the previous section we have the photon propagator1 µ k ν = −i k2 gµν 1 − 1− + iǫ λ (k 2 kµ kν . 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 .

s3 )γ µ u(k.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. 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 ).s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s4 )(−ie)γ ν u(p.s4 e4 (m) µν (M) L L .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . s4 )γµ u(p. To lowest order (∝ α = e2 /4π) only one diagram contributes. s2 )|2 ¯ q2 (10. s)γν u(k ′ . s1 ) ¯ −igµν u(p′ . s)¯(k. s′ ) ¯ u s.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. s3 )(−ie)γ µ u(k.79) . ¯ q2 (10.76) 1 . s1 ) u e2 u(p′ . 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 ′ .s2 . s′ )γµ u(k. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. t2 (10.s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ .5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon.Feynman diagrams for scattering amplitudes 100 10. where u ¯ L(m) µν = = 1 2 u(k ′ .77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10.s3 .

Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s3 )γ µ v(p′ . 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. (c) Do the same for the connected Green function Gc . To lowest order (∝ α = e2 /4π) only one diagram contributes.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 .s4 = 1 2s × v (k ′ . s3 ) ¯ ¯ 1 . s2 )γµ u(k.2.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ .s3 . The diagram. s4 )γ ν u(p. . . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s2 (10. Exercises Exercise 10. . the masses. This is known as crossing symmetry. x4 ) for the interacting case in φ4 -theory to ﬁrst order in the coupling constant g as obtained from the full expression for Z[J] in section 10. s4 ) u e2 v (k ′ . s3 )γ µ v(p′ . s4 ) v (p′ .s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. s1 )|2 ¯ s 1 .s2 3 . Similarly. (4) (4) . s2 ) ¯ ¯ 1 2s u(p. s2 )γµ u(k. s1 ) u(k.s2 . s1 )γν v(k ′ .

. Express the contributions in invariants s. in the amplitude (Tij = A∗ Aj ). . . their sum is zero.37 and 10.38. Exercise 10. + pn ). . Pictorially this implies. Show that i the amplitude is symmetric under the interchange of t ↔ u. . . .3 Show that the combinatorial factors found using the rules given in section 10. . It is of the form −i M = A1 − A2 . . . . hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . or if we in general would consider all momenta as incoming.4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. . but the incoming and outgoing momenta are identical. xn ) ∝ (2π)4 δ 4 (p1 + . . pn ). + pn ) G(n) (p1 . xn ). e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. xn ) ≡ (2π)4 δ 4 (p1 + .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. . which means overall momentum conservation in Green functions in momentum space. . . for instance. 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). xn + a) = G(n) (x1 . Exercise 10. . . . . . .2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. 1 2 (b) Calculate the quadratic pieces and interference terms. . Note that the function G(n) (p1 . |M |2 = T11 + T22 − T12 − T21 . . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . pn ) only has n − 1 independent variables in it. t and u. . . . .Feynman diagrams for scattering amplitudes 102 Exercise 10. .

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

use the ﬁrst relation and the abovementioned threshold value for s to calculate the threshold for a speciﬁc n-particle ﬁnal state in the target rest frame. for instance.8) (11.16) . m2 . m2 ) a b . 2mb λ(s. 4π s 4π 8π s 4π (11.11) (11.12) pb = (mb . 2mb (11.13) (11. Also in this case one can express the energy and momentum in the invariants.14) One can. p2 ) = CM = = 1 (2π)2 1 (2π)2 1 (2π)2 d3 p1 d3 p2 4 δ (P − p1 − p2 ) 2E1 2E2 √ d3 q δ( s − E1 − E2 ) 4 E1 E2 q 2 d|q| √ dΩ(ˆ) q δ( s − E1 − E2 ) 4 E1 E2 which using |q| d|q| = (E1 E2 /(E1 + E2 )) d(E1 + E2 ) gives dR(s. p1 . m2 ) = 4(pa · pb )2 − 4p2 p2 . (11. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . = 2 s λ(s. which in the speciﬁc case a b also can be expressed as λ(s. m2 . ptrf ). 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. In that case trf pa = (Ea . 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 √ = . p1 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 .15) Explicit calculation of the reduced two-body phase space element gives dR(s.9) (11. mb and s). Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . 2 s s − m2 + m2 a b √ . a (11. 0). m2 ) a b .10) The function λ(s. m2 ). m2 . m2 . m2 ) is a function symmetric in its three arguments. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . 4s (11. p2 ) = = where λ12 denotes λ(s.

20) → π0 + n → π+ + π− + n → . dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . usually is a 2-particle state.. E . A speciﬁc reaction channel starts contributing at the threshold value (Eq. py ). Instead of the scattering angle.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes.17) (11.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. in which one particle is detected. The variable s is always larger than a c b d the minimal value (ma + mb )2 . s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.4).. The initial state. 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 . the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . Consider the process a + b → c + d. 2s (11. tmax min λcd /4s. however. .21) (11.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. = (11. At high energies. usually many particles are produced. In inclusive measurements no particles are detected in the ﬁnal state. 11. The various possibilities are referred to as diﬀerent reaction channels. in exclusive measurements all particles are detected. Consider the 1-particle inclusive case. there is usually a preferred direction. pc .19) (11.27) with m2 = m2 − p2 = m2 + p2 + p2 . Examples appear in → π− + p (11.22) (11. Of course there are not only 2-particle channels. pT = (px . Writing E = mT cosh y.25) dR(s. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.18) (11. pz = mT sinh y. An often used set of invariants are the Mandelstam variables. (11. pd ) = 4π s 4π 8π s 4π dt dt √ √ . a c with q = λab /4s and q ′ = being 0 or 180 degrees. 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 . H1 + H2 → h + X.

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

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

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

8π s 2π 8π s . θ) in quantum mechanics. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . note the sign and cos θ = qi · qf ). The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. . 2 λab and for instance for two particles dσ = q ′ M (s. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ).41) (11.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . .47) M (s. q n ).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 . θcm ) √ q 8π s 2 2 (11. p1 . pn ) |M |2 q 32π 2 m2 dΩ |M |2 .42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. 11. (11. . t) λab 4π dt. . q f ). . i.45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM.Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). ˆ ˆ Inserted in the unitarity condition for M . 16π 2 s (11. . p1 . for which one has dσ/dΩ = |f (E. θ)|2 . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . respectively. (11. implies for the scattering matrix M . .e. The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. θ) = −8π s ℓ (in analogy with the expansion for f (E. (11. . (11.43) dΩcm = π M (s. . p1 . pn ). see 11. . pn ). q n ) q n qn √ Mni (q i . 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 . for which the √ amplitudes squared have been calculated in the previous chapter. .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. θ) or M (q i . θ). . (11. The result is N = |Mf i |2 dR(s.

Scattering theory 110 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i . the total cross section can be determined. (11. ℓ (11. n (11. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . (11.e.48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . (11. In general a given channel has |Sℓ (s)| ≤ 1.49) or Im Mℓ = q |Mℓ |2 . q ˆ i. † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni .52) From the imaginary part of M (s. Using σ= dΩ q ′ M (s. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)). in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. 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 . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . 0). (11. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ).54) ℓ .50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ).53) The diﬀerence is the inelastic cross section. θcm ) √ q 8π s 2 . 2i q 2i q (11.51) and for the case that this is the only channel (purely elastic scattering.

i. . For instance the tree-level calculation of Feynman diagrams at order e2 was real. This is what we will do to discuss decay widths.57) This shows that Γ is the width of the resonance. however. .e. which is given by a sum of the partial widths into the diﬀerent channels.e. . 0). ∝ U (t. such that |U (t.55) (note that we have disregarded spin). It is important to note that the physical width of a particle is the imaginary part of the two-point vertex at s = M 2 . σel = σinel . 0) ∝ e−i(E−iΓ/2)t . This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. (11. For an unstable particle one expects that U (t. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . The prescription for the pole structure. but still real. To check unitarity we need the amplitude at order e4 . 0)|2 = e−Γt .Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. For the amplitude in a scattering process going through a resonance. The quantity Γ is precisely the width for unstable particles. 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. 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 .5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. it is straightforward to write down the partial wave amplitude. . the time-evolution operator. s − M 2 + iM Γ (11. We note that unitarity is generally broken in a ﬁnite order calculation. in which case.58) . (q Mℓ )ij (s) = −M Γi Γj . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. 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 product of amplitudes is of order e4 . η = 0.56) (again disregarding spin). i. 11.

u) = g(t. which is attributed to neutrinos. the full width is Γ = 88 keV. t)} .26 keV. Limiting ourselves to a resonance in one channel.e.2 GeV. The cross section at resonance is about 30 nb. 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. At resonance the cross section σT (π + p) is about 210 mb. M2 − s (11.Scattering theory 112 (Prove this using the unitarity condition for partial waves). Γ = 2.88 MeV. 2 q (s − M 2 )2 + M 2 Γ2 (11. Its mass is M = 1232 MeV.59) reaching the unitarity limit for s = M 2 . t) + g(u.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 . it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. The half-width of the resonance is M Γ. Knowing that each neutrino contributes about 160 MeV (see next chapter). one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . Its mass is M = 3096. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. • The Z 0 resonance in e+ e− scattering with M = 91. where furthermore σinel = 0.49 GeV. Exercises Exercise 11. dt s(s − 4m2 ) with f (t. This characteristic shape of a resonance is called the Breit-Wigner shape.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). its width Γ = 120 MeV. i. Note that because of the presence of a background the phase shift at resonance may actually be shifted. the partial width into e+ e− is Γee = 5. a fast change in the phase shift for a narrow resonance. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. u) + f (u.4) can be written as 4πα2 dσ = {f (t. This is a narrow resonance discovered in 1974. u) + g(t. Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering.

6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude.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. As discussed for unstable particles. Thus.999 877 Γ and Γ(π − → e− νe ) = 0. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width.2).2 Calculate the ﬂux factor. Calculate now fπ .000 123 Γ. Exercise 11. ρcm ρcm a b 1 cm .Scattering theory 113 Exercise 11. Compare the results with the experimental π − lifetime τ = 2. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . a quantity which we will ¯ ¯ encounter in the next chapter (section 12.4 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∆ . Exercise 11. writing N (s) M1 (s) = 2 + iM Γ .8. 4. cm |va − vb | in the center of mass system for two colliding particles a and b. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.

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

i. ds which is solved by dψ(s) ds = dxµ . 12. but expresses them with respect to diﬀerent bases. In principle such a phase in a given point is not observable. = . should be independent of such transformations. (12.e. (12. i. Dν ] ψ(x) 1 − ig dσ µν Gµν 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.19) This gives rise to a (path dependent) phase in each point.15) ψ(x). which connects two identical vectors. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). (12. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. ds s dxµ ds′ P exp ig A (s′ ) ψ(0).16. x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). However. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.The standard model the basis for x G and x+dx 117 G. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x).14) (12. (12. The relation in Eq. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. 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 . -dx -dy µ µ dyµ µ dx For a vector. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12.e. 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).18) or considering a path xµ (s) from a ﬁxed origin (0) to point x.

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

23) F +η the latter only forming a minimum for m2 < 0. 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 . so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . for the vacuum Qa |0 = 0.22) F Realization of symmetries (model independent) In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. while the lagrangian including the nonzero groundstate expectation value chosen by nature. In the case of degeneracy. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic.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. i. (12.e. A quantum ﬁeld theory has only one nondegenerate groundstate |0 . Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld.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. It is only invariant under rotations around the 3-axis.25) 2 2 2 Therefore there are 2 massless scalar particles. The result is 1 1 1 1 (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 − m2 (ϕ2 + ϕ2 ) L = 1 2 2 2 2 2 1 2 1 − m (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12. . It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . ϕc = 0|φ |0 = (12. . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. . say 0 0 . while the quadratic terms must correspond with non-negative masses. etc. (12. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. The η massless particles are called Goldstone bosons. therefore a choice must be made. In this situation one speaks of spontaneous symmetry breaking. (12. The situation now is the following. They are known as the Weyl mode or the Goldstone mode: Weyl mode. has less symmetry. The classical groundstate appears degenerate. In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 .) terms constitute interaction terms. Any constant ﬁeld with ’length’ F is a possible groundstate. ϕ2 and η. In this case the spectrum is described .21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero).

As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. Including a sum over the diﬀerent ﬂavors (up.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. a and a′ are degenerate states. H] = 0. Using the Dirac equation. if the generators Qa generate a symmetry. etc. strange. Note that for the ﬁelds π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x .32) (12. A bit more formal. 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. ∂ ∂ (12. . there must be a nonzero expectation value 0|Jµ (x)|π (k) . i. . it is easy to see that one gets ∂µ V µ = i ψ [M. denoted |π a (k) . ψ −→ ei α·T ψ. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). . 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. ψu mu md ψ = ψd . (12.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. V µ = ψγ µ T ψ. The generators Qa (in that case the rotation operators Lz .27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. T ] ψ. = ψ(i/ − M )ψ. 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′ . [Qa . Taking the derivative. a ∂ µ Jµ = 0. T = τ /2. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6). i. (12.e.e. Neglecting at this point the local color symmetry. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). (12. Goldstone mode.31) . a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab .The standard model 120 by degenerate representations of the symmetry group.. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space.e. a massless Goldstone boson for each ’broken’ generator. . irrespective of the fact if they annihilate the vacuum. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. i. called ﬂavors.) one can write L = f ψ f (i/ − Mf )ψf . M = . π (12.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor).26) for all the states labeled by a corresponding to ’broken’ generators. and one must have mπa = 0. This means that they are operators that create states from the vacuum. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. down.

35 is nonzero. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry.34) Using the Dirac equation.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. much smaller than any of the other mesons or baryons.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. it is easy to see that one gets ∂µ Aµ = i ψ {M. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. proton and neutron. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . or. a doublet (isospin 1/2) of nucleons. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. for which the ﬁeld (according to the discussion above) has the same behavior under parity. There exists another set of symmetry transformations. e. From group theory (Schur’s theorem) one knows that the latter can only be true. (12. This combined symmetry is what one calls chiral symmetry. M = m · 1. etc. as the quantity ∂µ Aµ (x). nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . parity minus). socalled axial vector transformations. How can we see this. 12. generated by TR/L = 1 (1 ± γ5 )T . however. (leaving out the ﬂavor structure) the same as ψγ5 ψ. the lagrangian density is invariant under left (L) and right (R) rotations independently. which is approximately true for the up and down quarks. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . behaves as a pseudoscalar particle (spin zero. T ] = 0. (12. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. Still the basic fact that the generators acting on the vacuum give a nonzero result (i.e. This symmetry. In the real world.27. chiral symmetry is not perfect. i. while the groundstate is only invariant under isospin rotations (R = L). including in Dirac space a γ5 matrix.g. i. etc. i. This situation.e. From the number of broken generators it is clear that one expects three massless Goldstone bosons. which is equivalent 2 to the V-A symmetry. The chiral ﬁelds ψR and ψL are transformed into each other under parity. ∂ ∂ (12.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. Aµ = ψγ µ T γ5 ψ. with almost degenerate masses (Mp = 938.6 MeV/c2 ). We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved. 12.e. where the quark masses are not completely zero. parity would not play a role in the world. fπ = 0 remains. is what happens in the real world. Therefore the world of up and down quarks describing pions. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. I. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. ψ −→ ei α·T γ5 ψ. 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. ψR/L = 1 (1 ± γ5 )ψ. is by nature not realized in the Weyl mode.e.e. . (12.3 MeV/c2 and Mn = 939. stated equivalently. if in ﬂavor space M is proportional to the unit matrix. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . T } γ5 ψ. Note that these transformations also work on the spinor indices. T γ5 = τ γ5 /2. but also a triplet (isospin 1) of pions. both are very small. but the fact that the symmetry is not perfect and the right hand side of Eq.

having two independent spin components) and three scalar ﬁelds (one degree of freedom each). 4 2 where Since the explicit (adjoint. (12.40) a Dµ φ = ∂µ φ − ig Wµ La φ. made into a gauge theory. as in this case there are no massless Goldstone bosons. (12.37) (12.3 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. The symmetry is broken in the same way as before and the same choice for the vacuum. The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. is made. while the number of massive gauge bosons coincides with the number of ’broken’ generators. We have the possibility to perform local gauge transformations. 0 ϕc = 0|φ |0 = 0 . 0 0 0 = .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 + . Initially there are 3 massless gauge ﬁelds (each. in this case three-dimensional) representation reads (La )ij = −i ǫaij . 0 φ= (12.39) (12. again 9 degrees of freedom.e. . thus 9 independent degrees of freedom. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. i. Note that the number of massless gauge bosons (in this case one) coincides with the number of generators corresponding to the remaining symmetry (in this case rotations around the 3-axis).38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . like a photon. one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. .The standard model 122 12. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. After symmetry breaking the same number (as expected) comes out. but one has 1 massless gauge ﬁeld (2). The diﬀerence comes when we reparametrize the ﬁeld φ. Dµ φ = Gµν = ∂µ φ + g Wµ × φ. . (12. One may wonder about the degrees of freedom. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m .

while the lefthanded particles form a doublet. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. ∂ ∂ ∂ (12. 0 1 +1/2 L νe 3 − = with TW = and TW = L= −1/2 eL L 2 R = R − = eR 3 with TW = 0 and TW = 0.e.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.47) and YW is considered as an operator that generates a U (1)Y symmetry. ψR/L = 2 (1 ± γ5 )ψ decouple. i. 2 (12.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. explicitly L −→ e−i β/2 L. down. its corresponding neutrino (νe . 2 2 = ∂µ R + i g ′ Bµ R. the quarks (up.48) (12.) and the gauge bosons responsible for the strong. muon µ− or tau τ − ). For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. neutrinos. This is written as 3 Q = TW + YW . R −→ e−i β R. νµ and ντ ) and their antiparticles.45) (12. etc. (12.).and righthanded particles. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature.50) (12.51) .The standard model 123 12. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . left. electromagnetic and weak forces. that gives a good description of the physical world.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. (12.44) which has an SU (2)W symmetry under transformations eiα·TW . U(1)Y U(1)Y (12. etc. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R.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. muons. The procedure. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. the leptons (elektrons. τ ). The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. however. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. = ∂µ L − (12. R. 1 As they are massless. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L.

the U (1)Q symmetry (generated by the charge operator). .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . + 8 (12. † i i g Wµ · τ − g ′ Bµ )φ. leading to the parametrization 1 0 φ(x) = √ (12. 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). . .58) (12.52) L L (f 1) (f 2) = = i Rγ µ (∂µ + ig ′ Bµ )R + i Lγ µ (∂µ + i ′ i g Bµ − g Wµ · τ )L 2 2 1 1 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν )2 − (∂µ Bν − ∂ν Bµ )2 . this leads to the lagrangian L (h1) .The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1.56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . 4 4 In order to break the symmetry to the symmetry of the physical world. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. . 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. L where L (h1) (h) =L (h1) +L (h2) . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. (12. −V (φ) L and (h2) = −Ge (LφR + Rφ L).57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + .59) . 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. h) (12. For the classical ﬁeld the choice θ4 = v is made.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. (12. (12. . Using local gauge invariance θi for i = 1.

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 . g ′ /g = tan θW . 2θ 2θ 4 cos W cos W 0. 2 Zµ (12. (12. 2 (12. we can read oﬀ e = g sin θW = g ′ cos θW . 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 . 4 2 MW g2 v2 = . .66) From the coupling to the photon. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.67) The coupling of electrons or muons to their respective neutrinos.69) .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 . .61) (12. for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + .63) (12.68) 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 . (12.60) (12. and correspond to three massive particle ﬁelds (W ± and Z 0 ) and one massless ﬁeld (photon γ) with 2 MW 2 MZ 2 Mγ = = = g2 v2 .62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .64) (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.

Comparing this with the total width into (invisible!) channels. . With the choosen vacuum expectation value for the Higgs ﬁeld. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).80) First. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.73) (12.231 2.75) 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 . Finally we want to say something about the weak properties of the quarks. 2 2 2 (12.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. i. Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three. (12. one has the relation g2 e2 1 GF √ = . me /v is extremely small. g ′ and v determine a number of experimentally measurable quantities. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).e. (12. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. The coupling to the Higgs particle is weak as the value for v calculated e. 3 = TW . such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127.74) (12. 80.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.77) (12.e. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.72) (12. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .g. i. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).71) GeV −2 = = = = . giving rise to a running coupling constant after renormalization of the ﬁeld theory.166 4 × 10 0.5 MeV (exp.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 1. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions.76) with CV CA 3 = TW − 2 sin2 θW Q. −5 (12.40 GeV. v (12. 91. (12. from the MW mass is about 250 GeV.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. (12. muon or tau.19 GeV. g2 MZ 2 2 (CV + CA ).

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

allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md .85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . d d (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.g.82) space. 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 .The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. i. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . u u where we include now the three lefthanded quark doublets in QL . e. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . UR = uR cR tR . The Higgs ﬁeld is still limited to one complex doublet.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices.q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . (12.86) the unitary CKM-matrix introduced above in an ad hoc way. respectively). † Λu Λ† = Vu G2 Vu . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. ms and mb ).34.e.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.227. Working with the mass eigenstates one simply sees that the weak current coupling to the W ± mass † mass becomes U L γ µ DL = U L γ µ Vu Vd DL . the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . magnitudes of the entries in the CKM matrix are nicely represented parametrization 1 − 1 λ2 λ λ3 A(ρ − i η) 2 1 2 −λ 1− 2λ λ2 A V = 3 2 λ A(1 − ρ − i η) −λ A 1 12.83) where Vu and Vd are unitary matrices. A ≈ 0.22 and η ≈ 0. the weak mass eigenstates are L (h2. (12. each of these containing the three families. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues.82 and ρ ≈ 0. .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.

(12. (12. In this case. and obtain L (h2.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. there is no family mixing for massless neutrinos. decoupling from all known interactions. 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 . NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . As before.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). i. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . e N L= L EL (12.92) . The mass ﬁelds ER † mass † mass = We ER . since the weak current is the only place where they show up.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12.89) with three righthanded neutrinos added to the previous case.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. 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 . The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . (12.5×10−3 eV2 . EL = Ve† EL . UPMNS a parametrization that in principle also could have been used for quarks. 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. mµ and mτ . one can write Λe = Ve Ge We and we ﬁnd † L (h2.93) 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 .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. we ﬁnd massless neutrinos.90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .e. EL = Ve EL . † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn .ℓ) = −LφΛe ER − ER Λ† φ† L. (12.

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.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. 1 L Υ ′ = nR + nc .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. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors.ν =− containing three (CP-violating) phases (α1 . one being massive. are equivalent to two decoupled Majorana neutrinos (see below). The mass matrix can be written as ML |MD | eiφ M = (12. (12. (12. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. a unitary matrix relating them to the weak eigenstates. 12.99) .ν = − MR NR NR + MR NR NR . is to add in Eq. For these light Majorana neutrinos one has.and righthanded species then just form a Dirac fermion. hence the mixing matrix becomes iα /2 0 0 e 1 (12. 1 ∗ c c (12. Thus (disregarding family structure) one has two Majorana neutrinos. one must for the neutrinos as well as charged leptons. For the neutrino sector. coupled by a Dirac mass term. however.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. 0 0 1 L mass. We use here the primes starting with the weak eigenstates. 2 although this option is not attractive as it violates the electroweak symmetry.95) 2 In order to have more than a completely decoupled sector.The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. 1 c ∗ c L mass.98) (12. 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. Absorption of phases in the states is not possible for Majorana neutrinos. the massless and massive Majorana neutrinos. This is called the see-saw mechanism (outlined below). couple the right.c. Thus 2 Υ ′ = nc + nL . the left.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . as above. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. with for the righthanded sector a mass term MR .94) M L NL NL + M L NL NL . − MD MR nR 2 (12. 2 R ψ = nR + nL . For the leptons. Actually. α2 and δ).

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).The standard model 131 taking ML and MR real and non-negative. T } γ5 ψ. T ] ψ. H] = 0. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . Prove that if the generators Qa generate a symmetry. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. L Υ2L nR Υ1 and Υ2 .2 Derive for the vector and axial vector currents. The diagonalization is analogous to what was done for the Λ-matrices and one ﬁnds U M U T = M0 with a (unitary) matrix U . Υ2R nR (12.1 Consider the case of the Weyl mode for symmetries.105) for i = 1. 2 U (M M † ) U † = M0 . (12.3 (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. Exercise 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′ . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. a and a′ are degenerate states. i. .103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. i ψ {M. 2 with real masses Mi .101) Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2 . i. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR .e. (12.102) = 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 . related via c Υ1R = U nL . Exercises Exercise 12. (12.e.

at least up to a high (which?) order in λ. Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 . to be precise one has for bosons (b) or fermions (f). etc.The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . hhh. Exercise 12. Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . Γ(Z 0 → e+ e− ). mb ≈ 5 GeV and MW ≈ 80 GeV. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld.5 Show that the couplings to the Higgs (for W + W − h. (b) Calculate the width to electron-positron pair. 2 2 The masses are mt ≈ 175 GeV. Exercise 12. e+ e− h.231. In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) .7 In this exercise two limits are investigated for the two-Majorana case. and the width to a pair of neutrino’s. ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ).) are proportional to the mass squared for bosons or mass for fermions of the particles. Exercise 12.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. 3 TW . the weak mixing angle is given by ¯ sin2 θW = 0.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. Exercise 12. Mf M2 and ghf f = . ZZh. . The mass of the Z 0 is MZ = 91 GeV. Γ(Z 0 → νe νe ).

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

relativistic quantum mechanics

relativistic quantum mechanics

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

Are you sure?

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

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

Get the full title to continue

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

scribd