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

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

November 2011 (version 6.04)

Contents

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

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

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

1

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

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

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

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

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

These notes Section 1.H. C. Cambridge University Press. M. 2007.3 2. I: Foundations.2 3.2 36 22 . Peskin & Schroeder and Srednicki.1 2 2 3. 2. The schedule in the Fall of 2011 is Chapters 1 through 10 in period 2 (7 weeks in November and December 2011) and Chapters 11 and 12 in period 3 (January 2012).3 3.1 6.5 4.E. Zuber.2 6. 1995. Schroeder. Cambridge University Press.2 3.3 3. S.2 3.3 4. 1980.-B. Vol.2 2.2 1. Veltman.3 6.2 2. Peskin and D.5 5.1 2.2 3. Diagrammatica. Weinberg. Cambridge University Press. 6.4 2. Addison-Wesly. 5.3. 1996. Cambridge University Press. The quantum theory of ﬁelds. 1985. II: Modern Applications. Ryder. 4. Vol. M.3 References As most directly related books to these notes.2 4.V. Itzykson and J.3 3.3 3. The notes contain all essential information.3 2. Quantum Field Theory.4 3. An introduction to Quantum Field Theory. 1994. 3. M. L. Other text books of Quantum Field Theory that are useful are given in refs [3-6].7 2.1 4. 3.4 3. 1.4 3.1 5.1 2. 2. Quantum Field Theory. Cambridge University Press.2 6.5 2. Srednicki.3 3. 1995.4 4.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2.2 3. Quantum Field Theory. 3. but are rather compact. McGraw-Hill.2 2. I refer to the book of Srednicki [1] and Ryder [2].1 1.1 3.1.3 2. Corresponding chapters in books of Ryder.5 2.4 6.4 2.1 3.3 2.3 2.2.8 3.

5 4.5 9.2.1.4 9.7 4. 6.1 20. 6.1 5.2 12.3 7.2 9.6 6. 9.5 .8 6.2 5.2. 8.10 8.6 8. 2.3 12.3 20.4 10.6 10.6 3.1 20.1 7.5 11.2 10.5 4.1 11.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.3 8.5 6.5 7. 6.6 9. 6.1 9.3 7.2 7.2.4.3 10.1 20.2 8. 9.8 5 4.6 3.3.2 11.1.5 7.2. 2.4 4.5 2.2 4.4 12.5 9.7.2 8. 6.3.3 8.6 3. 20.7 5. 4.3 2.4 3.1.3.1 8.4 8.6 4.4 7.4 12.3 11.4.4.1 12.3 6 8 9 Srednicki 3 3 3 37 5. 4.4 2.1 10.4 These notes Section 7.5 6 7 6 6.5.3 3.1 9. 4.3 6. 4.3 9.

3 • First two sentences of Chapter 5 have been modiﬁed.5 Corrections 2011/2012) • Extension of Exercise 3. • Slight rewriting of discussion of Majorana mass term in section 6.2 .

the action of the momentum operator in the coordinate representation is not as simple as the position operator.5) One can also choose a representation in which the momenta of the particles are the dynamical variables.2) In the limit that v ≪ c one reaches the non-relativistic domain. ≡ h/2π = 1. They satisfy the well-known (canonical) operator commutation relations [ri . In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). c = 299 792 458 m s−1 . (1. S ≫ .1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. One can talk about states |r and the wave function ψ(r) = r||ψ . (1. In special relativity a special role is played by the velocity of light c. Indeed. usually given divided by 2π.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined. ˜ ψ(p) = i d3 r exp − p · r ψ(r). pj ] = i δij . 1 (1.4) where δij is the Kronecker δ function. In quantum mechanics a special role is played by Planck’s constant h.6) . (1. In quantum mechanics. quantum eﬀects do not play a role anymore and one is in the classical domain. (1. being eigenvalues of the position operators. 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.582 119 15 (56) × 10−22 MeV s.054 571 68 (18) × 10−34 J s = 6. (1.1) In the limit that the action S is much larger than . Corresponding position and momentum operators do not commute.Chapter 1 Introduction 1. It is given by pop ψ(r) = −i ∇ψ(r). the position can in principle be determined at any time with any accuracy. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions.

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.

such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 .Introduction 5 that can be used in the cross product of two vectors z = x × y. The Poincar´ transformations include e e Lorentz transformations and translations. Λµn n T ν1 . x1 . So we could have used all upper indices in the above equations. • ds2 < 0 (spacelike intervals). x1 . linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. No diﬀerence is made between upper or lower.1.22) (1.µn = Λµ1ν1 . the points lie on the lightcone and they can be connected by a light signal. In special relativity we start with a four-dimensional real vector space E(1. it is convenient to distinguish lower indices from upper indices. Unlike in Euclidean space. x) = (x0 . The real. linear transformations that do not change the length of a four-vector are called the Lorentz transformations. In Minkowski space. xi are the three space components. Many other four vectors and tensors transforming like T ′ µ1 . Vectors are denoted x = xµ nµ . The length (squared) of a vector is obtained from the scalar ˆ product.24) The quantity gµν ≡ nµ · nν is the metric tensor.. the invariant length or distance (squared) is not positive deﬁnite. denoted as V ′µ = Λµν V ν .. and are given by (x0 . x3 ) = (t. given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). (1. ˆ ˆ (1. . are . We note that for Euclidean spaces (with a positive deﬁnite metric) vectors and tensors there is only one type of indices.26) (1. One can distinguish: • ds2 > 0 (timelike intervals). 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 .23) δjm δkn − δjn δkm .25) Within Minkowski space the real. however. x2 .3). one must distinguish tensors with upper or lower indices and one can have mixed tensors.21) (1. x3 ). Because of the diﬀerent signs occurring in gµν . • ds2 = 0 (lightlike or null intervals). Useful relations are ǫijk ǫℓmn ǫijk ǫimn ǫijk ǫijl = = = δiℓ δjℓ δkℓ 2 δkl . where the coordinate t = x0 is referred to as the time component. Relations relating tensor expressions. For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. When 3-dimensional space is considered as part of Minkowski space. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ .. The (invariant) lengths often have special names. The lower indices are constructed in the following way.. One has x2 = xµ xµ = t2 − x2 . x2 .2.νn can be conν structed. 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. These transformations do change the components of a vector. in which case zi = ǫijk xj yk . δim δjm δkm δin δjn δkn . we will use upper indices for the three-vectors. . xµ = gµν xν . x2 = x · x = xµ xν nµ · nν = xµ xν gµν .3) with basis nµ (µ = ˆ 0. independent of a coordinate system. (1. 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. −x). 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 .

Each of these permutations leads to a minus sign. The length squared of ∂ is the d’Alembertian operator. deﬁned by ∂2 − ∇2 . The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . starting from ǫ0123 = 1. (1. (1.33) (1. = ∂ .27) It is easy to convince oneself of the nature of the indices in the above equation. g0 = g1 = g2 = g3 = 1.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . 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. Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ .31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ .35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ .36) Exercises Exercise 1. ∂2 . for instance. . not aﬀected by Lorentz transformations. (1.32) . Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α. ′ ′ .28) ν ν Note that gµ with one upper and lower index. where m is called the mass of the system. (1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. Note that in this equation one has on left. (1.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. constructed via the metric tensor itself.∇ . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . . a∞ = 4πǫ0 2 /me e2 .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. = −24.34) (1. but more important has the same eﬀect on lefthandside and righthandside. ∂3 ) = ∂ ∂ ∂ ∂ . ∂1 . because one has (1. . It is an invariant tensor. explicitly written as (p0 .30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used). ∂t (1. The product of two epsilon tensors is given by g µµ ′ g νµ ′ g ρµ ′ g σµ g νν ′ g ρν ′ g σν ′ ′ ′ ǫµνρσ ǫµ ν ′ ′ ′ ρ σ′ = − g µν ′ g νν ′ g ρν ′ g σν g νρ ′ g ρρ ′ g σρ ′ ′ ′ g µρ ′ g νρ ′ g ρρ ′ g σρ g νσ ′ g ρσ ′ g σσ ′ ′ ′ g µσ ′ g νσ ′ g ρσ ′ g σσ . p).Introduction 6 called covariant. p) = (E.

1. α and me c2 = 0. 0. Exercise 1. but only appropriate combinations. ν. a+ . 0. 1. 0) ˆ n0 = (1.3. (d) Use the value of the gravitational constant GN / c5 = 6.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). . . n2 . 1.(A · B) C using the properties of the tensor ǫijk given in section 1. a2 ] or [a− . Compare it with the proton mass and use Eq. n3 .511 MeV. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2. 0.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). a+ . .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . 1) ˆ . by a simple few-line reasoning [For instance: If {µ. a1 . 0. n1 . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . ]. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . (b) The coordinates (a0 . a1 . 3} the index α can only be equal to one of the indices in ǫµνρσ .3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . Exercise 1. e. [Note: what is the MKS unit for V/T?] Exercise 1. 0. a1 . n1 . ˆ 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). σ} is a permutation of {0. a2 ] we can ﬁnd basis vectors n− . Also for the coordinates [a− . g−− . a2 . 0. n+ . How are a+ and a− related to a+ and a− . This demonstrates how a careful use of units can save a lot of work. g+− and g−+ . a+ . re = e2 /4πǫ0 me c2 to the Compton wavelength.13 to give its actual value in kg. 2. ρ. Also construct and calculate the Planck length Lpl . ˆ ˆ ˆ ˆ These are n1 = (0. (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . One does not need to know . (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. which is the Compton wavelength for the Planck mass. 0) ˆ n3 = (0. me . 1. ǫ0 . n2 .2 Prove the identity A × (B × C) = (A · C) B . c. 0) ˆ n2 = (0. a3 ) are the expansion coeﬃcients using the basis vectors n0 .

2) Acting on the wave function one ﬁnds for a free particle. (2.3 are not covariant. 2 + M 2 φ(r. i ∂ ∇2 ψ(r. written as pµ −→ i∂µ is covariant (the same in every frame of reference). 8 .7) k 0 = ± k2 + M 2 = ±Ek . one obtains (2. t) = − ψ(r. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. φk (r. (2. ∂t 2M (2. ∂t p −→ pop = −i∇. ∂t2 (2. E = p2 . t) = ∂2 − ∇2 + M 2 φ(r. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. the interpretation of this equation as a single-particle equation in which φ is a complex wave function poses problems because the energy spectrum is not bounded from below and the probability is not positive deﬁnite.2.Chapter 2 Relativistic wave equations 2.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . namely plane waves characterized by a wave number k. (2.3) Equations 2.1 and 2.1 The Klein-Gordon equation In this chapter. 2M (2. t) = exp(−i k 0 t + i k · r). But the replacement 2. p2 = pµ pµ = E 2 − p2 = M 2 . t).5) with (k 0 )2 = k2 + M 2 .4) where M is the particle mass. p). t) = 0.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. Although it is straightforward to ﬁnd the solutions of this equation.

10) ∂t which follows directly from the Schr¨dinger equation. k) ≡ a(k) and φ(−Ek . And. we want the most general solution.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. just as the dependence on time was in ordinary quantum mechanics. having the KG equation as a space-time symmetric (classical) equation. for which we need the interpretation of φ itself as an operator. but rather the zero component of a four vector. relativistically the density is not a scalar quantity. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. (2π)3 2Ek (2. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. This leads to the existence of negative densities. rather than as a wave function. which is what Lorentz transformations do. (2. 2.14) ˜ ˜ Introducing φ(Ek .Relativistic wave equations 9 i.e.15) .11) Therefore.3). −i φ∗ ∇ φ .13) ˜ with (in principle complex) coeﬃcients φ(k). The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq. The KG equation. k) + ei k·x φ(−Ek .9) They satisfy the continuity equation. 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. j) = ↔ i φ∗ ∂0 φ. Then. This continuity equation can be written down o covariantly using the components (ρ. As we will see later both problems are related and have to do with the existence of particles and antiparticles. 2.5). −k) . there are no longer fundamental objections to mix up space and time. ∂µ j µ = 0. (2. This operator has all possible solutions in it multiplied with creation (and annihilation) operators.8) (2. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time. ↔ ↔ (2. ∂ρ = −∇ · j.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds. 2M 2M (2. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. there are solutions with negative energy.2) is j µ = i φ∗ ∂ µ φ or (ρ. however. j) of the four-current j. (2. The appropriate current corresponding to the KG equation (see Excercise 2. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2.

changing the sign of the spatial coordinates. 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) . It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0. after restricting the energies to be positive).21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . x ∂µ ∂ µ + M 2 φ(˜) = 0. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). (2. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). Performing some Lorentz transformation x → x′ = Λx. 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).16) The consequence of this is discussed in Exercise 2.3 Symmetries of the Klein-Gordon equation We arrived at the Klein-Gordon equation by constructing a covariant operator (∂µ ∂ µ + M 2 ) acting on a complex function φ. φP (x) ≡ φ(˜) (P for parity). (2π)3 2Ek (2. i.6 We will explicitly discuss the example of a discrete symmetry. for which we consider space inversion. Since a · b = a · ˜ it is easy to see that ˜ b. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). Another symmetry is found by complex conjugating the KG equation. (2.18) (2. 2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). 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) .Relativistic wave equations 10 In Eqs 2.17) (2. (2π)3 2Ek (2. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. −x) ≡ (˜µ ). This shows how parity transforms k-modes into −k modes.14 and 2. which implies (xµ ) = (t.19) implying that for each solution φ(x) there exists a corresponding solution with the same energy.23) . x) → (t. (2π)3 2Ek (2.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. (2.e.

Exercise 2. Note that a(k) and b(k) are independent of t. QV ≡ satisﬁes ˙ QV = − ds · j. t). Q = 0. S V d3 x j 0 (x). t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k.3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . for the mode amplitudes aC (k) = b(k) and bC (k) = a(k).15. (2π)3 2Ek where Ek = Exercise 2.4 ˜ Write down the 3-dimensional Fourier transform φ(k. Exercises Exercise 2. For the real ﬁeld one has aC (k) = a(k).1 Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume. t) = ei Ek t (i∂0 + Ek ) φ(k. t). ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ).Relativistic wave equations 11 i.e. k). t) ≡ d3 x φ(x. This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. |f ′ (xn )| . d3 k F (Ek .2 Show that if φ and φ∗ are solutions of the KG equation. 2. ↔ ↔ Exercise 2. that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). is conserved. letting V → ∞. k) = (2π)4 k2 + M 2 . t) for the mode expansion in Eq. ˙ and thus for any normalized solution the full ‘charge’.

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

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

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. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. ] that satisﬁes • ∀ x. (3. π) (n. They satisfy the commutation relations [Li . z]] + [y.4) R . n) = 1 cos ˆ 1 Note ϕ ϕ ˆ + i (σ · n) sin 2 2 = exp i ϕ ˆ σ·n . for rotations around z for instance. y] = −[y. z ) = lim N →∞ that are generated by i 0 . with real a’s and 3 (ai )2 = 1. x] = 0 (thus [x. [z. 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 . [x. x]). With matrix commutation this algebra satisﬁes the requirements for a Lie algebra.1: the parameter spaces of SO(3) (left) and SU (2) (right).2 π) (-n.6) Summarizing.Groups and their representations 14 (n. Ly and Lz . 2 (3. we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices. 0 (3. thus is as the surface i=0 of a sphere in 4 Euclidean dimensions. . Next. One way of viewing the parameter space. x]] + [z. These matrices can be deﬁned as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). It is straightforward to check that any (ﬁnite) rotation can be obtained from a combination of inﬁnitesimal rotations. is a 3-dimensional Lie-group.z N N or (Lk )ij = −i ǫijk . y] ∈ A. namely that there exists a bilinear product [. In the same way we can consider rotations around the x.8) ˆ that (σ · a)(σ · b) = a · b + i σ · (a × b). ˆ therefore. and thus (σ · n)2 = 1.7) = a0 0 −1 i 0 1 0 0 1 • [x. • [x. [y. ˆ R(ϕ. Locally this is a 3-dimensional Euclidean space and SU (2). These generators form a threedimensional Lie algebra SO(3).and y-axes 0 0 0 0 0 0 0 −i . which reﬂects itself in the noncommutation of the generators. y ∈ A ⇒ [x. Lj ] = i ǫijk Lk . Writing a0 = cos(ϕ/2) and a = n sin(ϕ/2) we have1 A = A(ϕ. (3.5) Rotations in general do not commute. y]] = 0 (Jacobi identity).

n) −→ R(ϕ. (3. for ﬁxed n. however. There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(ϕ. H] = 0 for g ∈ G. [g. immediately sees that the Lie algebras are identical. thus. since any ﬁnite transformation can be constructed from the inﬁnitesimal ones. .11) Thus σx /2.13) 0≤ϕ≤π The relation A(σ · a)A−1 = σ · RA a (valid for any vector a) can be used to establish the homomorphism (Check that it satisﬁes the requirements of a homomorphism. n! 2! n=0 ∞ (3. The inﬁnitesimal generators of SU (2) are ˆ obtained by considering inﬁnitesimal transformations. .12) One. n) ˆ −→ R(2π − ϕ. . For instance in a function space translations. now with radius 2π and with all points at the surface identiﬁed (see ﬁgure). it is suﬃcient that the generators commute with H.9) The parameter-space of SU (2). . σy /2 and σz /2 form the basis of the Lie-algebra SU (2). n) π ≤ ϕ ≤ 2π. (3. .Groups and their representations 15 where we have used the (for operators new!) deﬁnition eA ≡ An 1 = 1 + A + A2 + . H] = 0 for g ∈ G.15) [g. i. the above mapping corresponds to the trivial mapping of the Lie algebras.e.e.10) σ ˆ · n. 3. . 2! . Suppose we have a system described by a Hamiltonian H. one has a Lie algebra isomorphism that is linear and preserves the bilinear product.2 Representations of symmetry groups The presence of symmetries simpliﬁes the description of a physical system and is at the heart of physics. 2 (3.e. i. in particular that AB → RAB = RA RB ). The next issue is to ﬁnd the appropriate symmetry operators in the Hilbert space. In the full parameter space. with ˆ J ·n ≡ ˆ 1 ∂A(ϕ. 2 2 2 (3. SU (2) ≃ SO(3). the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. i. n) ≈ 1 + i ϕ J · n. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian.14) For a Lie group. They satisfy σi σj σk = i ǫijk . φ′ (r) = φ(r + a) = φ(r) + a · ∇ φ + 1 (a · ∇)2 φ + . thus. sometimes in more than one way (but this will be discussed later). Near the identity. n) i ∂ϕ = ϕ=0 (3. (3. also can be considered as a ﬁlled 3-sphere. ˆ ˆ A(ϕ. .

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

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

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 . e To derive some of the properties of the Lorentz group.g. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1.Groups and their representations 18 3. 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. 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. det Λ = −1 is called improper). (3. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.26) (3. which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . G = 10 (3. In this section we consider the Poincar´ group. (ii) |Λ00 | ≥ 1.21) 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. From this property. σ/2 in a two-dimensional (spin 1/2) case. (det Λ = +1 is called proper. e.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .27) (3.3 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. it is convenient to use a vector notation for the points in Minkowski space. Writing x as a column vector and the metric tensor in matrix form.22) (Note that xT is a row vector). consisting of the Lorentz group and translations.25) (3. Using (i) and (ii) the Lorentz transformations can be divided into 4 classes (with disconnected parameter spaces) det Λ +1 +1 -1 -1 Λ00 ≥1 ≤ -1 ≥1 ≤ -1 (3.

Groups and their representations = i=1 (Λ0 )2 . z ) = exp(i ϕ J 3 ). inﬁnitesimally given by ˆ ΛR (ϕ. 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 . it is easy to ﬁnd the following typical examples from each of the four classes. z ) ≈ I + i ϕ J 3 .32) 0 − 0 −1 0 0 0 0 −1 . K 3 ). K j ] = −i ǫijk J k . Returning to the global group. K j ] = i ǫijk K k . since the generators K do not form a closed algebra. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η.29) 0 0 1 0 3′ 3 V sinh η 0 0 cosh η V i 0 0 0 Note that Lorentz transformations generated from the identity must belong to L↑ . i proof: use ΛT GΛ = G and ΛGΛT = G. J 3 ). inﬁnitesimally given by ΛB (η. [J i . one distinguishes rotations 0 + and boosts. [K i . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. J j ] = i ǫijk J k . which satisfy the commutation relations (check!) [J i . 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. Using these explicit transformations. z ) = exp(−i ηK 3 ). (3. J = (J 1 . 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 = (3.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3. βγ = sinh η.28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. Thus 0′ 0 V cosh η 0 0 sinh η V 0 0 0 i 1′ 1 V V 0 0 0 0 0 1 0 0 2′ = 2 −→ K3 = V V 0 0 0 0 . In L↑ . Note that the velocity β = v = v/c and the Lorentz contraction factor γ = (1 − β 2 )−1/2 corresponding to the boost are related to η as γ = cosh η. z ) ≈ ˆ ˆ I − i η K 3 . The commutator of two boosts in diﬀerent directions (e. (3.g. K 2 . J 2 . The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. we have found the generators of rotations. since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. 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). Rotations around the z-axis are given by ΛR (ϕ. This is the origin of the Thomas precession. K = (K 1 .

inﬁnitesimally approximated by (I + ω. inf (3.4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3. [M µν . M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) . P+ etc. We will just require Poincar´ invariance for the generators themselves.34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0.38) Explicitly. (3. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). ǫ). i i0 i i0 1 ijk M jk 2ǫ (3. Of the four parts only L↑ forms a group. (3.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 .39) (e. In order to ﬁnd the commutation relations including the translation generators. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations. 3 30 with η = ω and K = M (e.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. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 . the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations.33) These four transformations form the Vierer group (group of Klein).36) (3. The extension to the Poincar´ group is e ↑ straightforward. The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . three of which only involve spatial coordinates (rotations) and three others involving time components (boosts).g. One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . Also this group can be divided into four parts. We therefore ﬁnd (again) that there are six generators for the Lorentz group. K = M ). 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. we continue using covariance arguments. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . a). This is a + normal subgroup and the factor group is the Vierer group. 3. There are six generators.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 .g. Summarizing.

known from non-relativistic quantum mechanics are obtained. a)P µ U † (Λ. [J i . .38 could have been obtained in the same way.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. P j ] = i ǫijk P k . 3. K j ] = −i ǫijk J k /c2 . Λµν pν ⇒ ωµν pν ⇒ ωµν = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . = −i (g µρ P ν − g νρ P µ ) . To ﬁnd other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant. the generators J± in the case of the rotation group. the tensor M µν in terms of boosts cK i = M i0 and rotations J i = 1 ǫijk M jk . P ρ ] = 0.44) (3. [K i . (3. H] = i P i . [J i . ǫ) = 1 − ωαβ M αβ + i ǫα P α . other states in that representation are obtained by the action of operators outside the maximal commuting set. P j ] = i δ ij H/c2 . a)P µ U (Λ. These deﬁne states with speciﬁed quantum numbers that are eigenvalues of these generators.47) (3. (3. (3. 3. ν (3. the generators J 2 and J 3 in the case of the rotation group. The eigenvalues of these will be the four-momentum pµ of the state. .41) 2 At this point. a) = Λ−1 µ P ν . [K i . K j ] = i ǫijk K k . We have here reinstated c.43) giving the following commutation relations by equating the coeﬃcients of ǫµ and ωµν . Of the generators of the Poincar´ group we choose ﬁrst of all the generators P µ . These form a group called the little group associated with that four vector. P ν ] [M µν . because one then sees that by letting c → ∞ the commutation relations of the Galilei group. The generators of the translations are the (momentum) operators P µ . .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. (3. H] = 0. . P j ] = [P i . [J i . 2 2 (3. [P µ . . = pµ |p. Inﬁnitesimally. For instance. Taking any one of the states in an irreducible representation. check that U1 U2 corresponds with Λ1 Λ2 ). . For instance. writing the generator P = (H/c. In order to label the states in an irreducible representation we construct a maximal set of commuting operators. as part of the set. that commute e among themselves.42) (to decide on where the inverse is. J j ] = i ǫijk J k . Explicitly.45) Note that the commutation relations among the generators M µν in Eq. P µ |p. leads to U (Λ. the generators M αβ no longer exclusively act in Minkowski space.48) . .46) [K i . (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . one obtains 2 [P i . P ) in terms of the Hamiltonian and the three-momentum operators. They just state that M µν transforms as a tensor with two Lorentz indices. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the deﬁning four-dimensional representation). H] = [J i . a) = Λµν P ν or U † (Λ.

52) (3. Pν ] = [Wµ . i ǫµνρσ W ρ P σ . Wν ] = −i (gµα Wν − gνα Wµ ) . . = = i exp − ǫµνρσ M µν pρ sσ |p. 2 (3.54) This will enable us to pick a suitable commuting ’spin’ operator. . 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.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. . . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. p0 > 0. 0. 0. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . (3. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . 2 exp (−isµ W µ ) |p. . Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p). the third case represents the vacuum. . (3.55) commute with all generators and therefore are invariants under Poincar´ transformations.49) (3.56) . . Wα ] = [Wµ . These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. M ǫijk .53) (3. which can thus be chosen spacelike.50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . . 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. while the others have no physical signiﬁcance. .51) The following properties follow from the fact that W µ is a four-vector by construction.

Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). p. i (3. |p3 |). n1 (p). W i (p) = −W · ni (p)).64) such that in an arbitrary frame where the four vectors p. 0. . 1 2 (3. The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions).e. (1. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). 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 .60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . 1. ms . 1. M2 (3. 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 = 0.65) . M2 (3. (3. 0. n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p).. Together with the four momentum states thus can be labeled as |M. The commutation relations [W i (p). (0. 3. 0). M J) with components W i = (M/2)ǫijk M jk .57) (i. W j (p)] = i M ǫijk W k (p). the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators.61) one sees that S i (p) i =− W2 . In that case the vectors ni (p) are just the space directions and W = (0. 1. 0). Having made a covariant decomposition. . −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. 0.63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). s.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.e. 0. 0. show that W /M can be interpreted as the intrinsic spin. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq.59) (3. . 0. (3. p0 > 0. 0). 2 .58) i.Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. (0.

1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq.67) (3. i W 2 (p). 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 . Thus 2 λ(p) = J ·p . The algebra of W 1 . This state is in physical applications nondegenerate and is denoted by |0 . |p. however. σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a).13.67 has a vanishing right hand side). (3. Exercises Exercise 3. But if M = 0 one has a diﬀerent algebra (Eq.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. implying continuous values for the spin (actually for W i = M S i ). (3. which can take any integer or half-integer value.70) as a commuting set with zero eigenvalues for P 2 and W 2 . The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values.73) (3.66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). 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). 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. Wν ]. W 2 and λ is derived from the general commutation relations for [Wµ . |p| (3. where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). Thus we have P 2 . λ . is characterized by its momentum and the helicity. a)|0 = |0 .68) (3. W 2 (p)] [λ(p). W 2 (p)] = = = i M 2 λ(p). W µ (p) = λ(p) P µ . −i W 1 (p). and gives [W 1 (p).Groups and their representations The above does actually include the massive case.71) which is called the helicity. We don’t know of any physical relevance. W 2 (3. 1 2 24 (3.69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). Note that angular momentum does not contribute to helicity as L · p = 0. thus. W 1 (p)] [λ(p). This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). 3. P µ . 3. The vacuum: pµ = 0. 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). A massless particle.

For a spin 1/2 antiparticle the representation A∗ is needed. φ0 ) doublet representation for SU (2)weak . Show that if det(U ) = 1. Examples are the quark (u. (b) This establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle. d) doublet representation for SU (2)isospin or the Higgs (φ+ . the (non-strange) η-meson state. This is true for SU (2) but it is. (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). Both particle and antiparticle have ’spin 1/2’. not true for SU (3) (ﬂavor or color properties of particles know two distinct deﬁning 3-dimensional representation. Exercise 3. one has to explicitly include this rotation. triplet and anti-triplet representation. Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and thus Tr(σi σj ) = 2 δij .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. . The appropriate isospin doublet to be used for antiquarks then is (−d. for instance.2 Show that for a unitary operator U = exp[iαk Jk ] with real coeﬃcients αk . . Show that the Lie-algebra representations J and J c are equivalent. and calculate U (a) p U −1 (a) . respectively). the trace of the operator Jk is zero (Convince yourself that det(eA ) = eTr A by diagonalizing A. Writing A∗ = exp(i α · J c ). Now write down the three isospin 1 meson quark-antiquark (meson) states. u).Groups and their representations 25 for any vectors a and b. (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. what are the matrices J c in the Lie-algebra representation describing the spin rotations for two-component ’anti-spinors’. In both of these cases the symmetry is broken by the charges ¯¯ of the particles. (b) Show that this transformation for the position operator in quantum mechanics indeed gives U (a) r U −1 (a) = exp (i a · p) r exp (−i a · p) = r + a. the operators Jk must be hermitean. Excercise 3. show that the transformation for the operators is given by O′ = U O U −1 = O + i [J. Exercise 3.3 (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. i.e. Is this easy to understand. O] + . Tr(σi σj σk ) = 2i ǫijk . the pions and the isospin 0 meson state. show that one matrix ǫ exist such that J c = ǫ J ǫ† . Note the diﬀerence with the (usual) spin 0 and spin 1 states composed from two spin 1/2 states. Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) . . Show actually that ǫ = 2 eiπJ (up to a phase).

From either of these. pj ] = [ri . you might just check relations involving J or K by taking some (relevant) explicit components. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane.69 of the Pauli-Lubanski operators for massless particles. derive the commutation relations and show that the algebra is isomorphic to (i. This is possible for a single particle.e.e.7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. if you don’t want to do this in general. p×s 1 (rH + Hr) − t p + . in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. Do this by showing that the set of operators. sj ] = 0. the operator U (u) that gives U (u) r U −1 (u) = r − u t. momentum and spin operators e satisfy the canonical commutation relations. they describe r −→ r′ = r − u t while t′ = t. For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. f (p)] = i ∇p f (p). = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). although that in this case is a relation that is found already from the Ehrenfest relation for O = ri .. i. (a) In classical mechanics. but it can be done. [ri . [ri . i.5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame. Exercise 3.6 Consider rotations and translations in a plane (two-dimensional Euclidean space). Exercise 3. Comment: extending this to more particles is a highly non-trivial procedure. Determine the commutation relations for the corresponding operators Ki in quantum mechanics. . pj ] = iδ ij and [si . (b) In quantum mechanics one has the Ehrenfest relation i dO = [O. (c) Give the unitary operator for boosts. and U (u) p U −1 (u) = p − m u. show ﬁrst the operator identity [r.Groups and their representations 26 Exercise 3.67 to 3. = r × p + s. Hint: for the Hamiltonian. = p. although the presence of an interaction term V (r1 . the others vanish. r2 ) inevitably leads to interaction terms in the boost operators. there is a one to one mapping) the algebra in Eqs 3. sj ] = iǫijk sk . indeed assures d K /dt = 0. H] + i dt ∂O . sj ] = [pi . consistent with the (canonical) commutation relations of a quantum theory. H P J K = p 2 c 2 + m2 c4 . boost invariance leads to a conserved quantity. that’s why many-particle non-relativistic quantum mechanics is ’easy’. These do not matter in the non-relativistic limit (c → ∞). rj ] = [pi .e.

U (ϕ) H(η) (4. 2 2 (4. 2 1 (J − i K). They will be labeled by two angular momenta corresponding to the A and B groups. Aj ] = [B i . C) is the group of complex 2 × 2 matrices with determinant one.8) ˆ ˆ ˆ with η = η n and ϕ = ϕn. This tells us how to ﬁnd the representations of the group. K = i J (A = 0). (4.4) (4. B j ] = i ǫijk Ak .1 The Lorentz group and SL(2.3) (4. C) can be written as a product of a unitary + matrix U and a hermitean matrix H. Special cases are the following representations: Type I : (j. K = +i . respectively.1) (4. similarly as the homeomorphism between SO(3) and SU (2). K = −i . The group SL(2. i ǫijk B k . j) K = −i J (B = 0). it also follows directly that the Lorentz group is homeomorphic with the group SL(2. With this choice of parameter-spaces the plus and minus signs are actually relevant. The matrices in SL(2. 0) Type II : (0. C).6) (4.2) 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 .10) . 2 2 σ σ J = .5) = 1 (J + i K).9) (4. They precisely correspond to the two types of representations that we have seen before. 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. 0. j ′ ). M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . (4. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. (j. B j ] = [Ai . 2 (4.Chapter 4 The Dirac equation 4. It is simply connected and forms the covering group of L↑ .7) From the considerations above.

C) and. thus. ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. one can use a boost to the frame with momentum ˆ ˆ p. cannot be connected by a Lorentz transformation belonging to L↑ . η → U η. H(η) = exp(−η · σ/2). The Lorentz transformations Λ(M ) and Λ(M ). ξ −→ Hξ. i. (4.18) As already discussed for SU (2). C) or L↑ . there exists the matrix ǫ = iσ 2 . however. −σ). Λ(ǫ) ∈ L+ . U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. (4.15) (4. C).The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. thus. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). (ǫ∗ = ǫ. σ µ ≡ (1. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 .13) 2M (E + M ) (exercise 4. ǫ−1 = ǫ† = ǫT = −ǫ). This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. (4. the ﬁrst equation shows that the conjugate representation with spinors χ∗ transforming according to −σ∗ /2 is equivalent with the ordinary two-dimensional representation (±ǫχ∗ transforms according to σ/2). (4. With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. The following relations for + rotations and boosts are useful. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . ˜ = σ µ Λ B aµ . while ±ǫη ∗ transforms as a type-I + spinor. U σ µ aµ U −1 = σ µ ΛR aµ . One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . ˜ (4. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. namely H(η) = exp(η · σ/2).17) that can be used to relate σ µ∗ = ǫ† σ µ ǫ = ǫ−1 σ µ ǫ and hence ˜ ¯ U ∗ = ǫ† U ǫ = ǫ† U ǫ.12) ξ → U ξ.e. η → Hη. This shows that M ∗ ≃ M . They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that y + ↑ ∗ ǫ ∈ SL(2.19) . if there does not exist a unitary matrix S. σ).18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. H ≡ (H † )−1 = H −1 ). This is not true for the two-component spinors in SL(2.16) = σ µ Λ R aµ . (4.1). C). such that M = SM S −1 .11) Considering ξ and η as spin states in the rest-frame. Eqs 4.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. In fact.

2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0. a) and ˜ a = (a0 . Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). A parity transformation changes aµ into aµ . m) = η(0. the boost in Eq. 0 χm uR H(p) .The Dirac equation 29 where The matrix I2 does not belong to L↑ . (vector). as we have seen in section 4. 4.23) u(p.22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. (pseudoscalar). Although these operators cannot be used to label the representations. = (4. m) = χm . For the spin 1/2 representations of the Poincar´ group including parity we.e. The i angular momentum operators J are represented by σ i /2. i. 4. they can be connected. ǫµνρσ will change sign. The behavior is the same for classical quantities.e. e. (scalar). the well-known two-component spinor for a spin 1/2 particle. C) are suitable for representing spin 1/2 particles. −→ −p (vector). and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. Thus M ∗ and M are not equivalent within SL(2. they cannot be connected by a unitary 2 2 transformation. we obtain for the two components of u which we will refer to as chiral right and chiral left components. For a particle at rest only angular momentum is important and we can choose ξ(0. generators. We also have seen that the representations (0. (axial vector). From the deﬁnition of the representations (0. the aij elements of a tensor will not change sign. This provides the easiest way of seeing ˜ what is happening.g. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . Taking M (Λ) = H(p). C). m) = χm uL H(p) 0 . 2 ) and ( 2 . −→ −r (vector). (4. but to L↑ . 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. must combine the repe resentations. Within the Lorentz group. we have seen that in the rest frame W i (p)/M → J i . 1 ) and ( 1 .1. that have the correct commutation relations. where a = (a0 .17. etc. but by a transformation belonging ↑ ↑ to the class P− .20) 4. 0). ) −→ ( . −→ M (Λ) (4.21) In nature parity turns (often) out to be a good quantum number for elementary particle states. The representing member of the class P− is the parity or space inversion operator Is . 2 2 (4. i. therefore. −a). 0) are inequivalent. + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . 0) of SL(2. 1 ) and ( 1 . It has the same eﬀect as lowering the indices.13. but rather the operators W i (p) should be used.

(iγ µ ∂µ − M ) ψ(x) = 0. It is of a form that we also played with in Exercise 2. = (4.28) which is a covariant (linear) ﬁrst order diﬀerential equation. We will discuss another explicit realization of this algebra in the next section.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . (4. 0 H 2 (p) uR uR .27 is known as the Weyl representation. . ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. From this one obtains the Cliﬀord algebra for the Dirac matrices. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . The general requirements for the γ matrices are thus easily obtained. (4. (4.33) deﬁne for a four vector a the contraction a = aµ γµ / . The explicit realization appearing in Eq. Applying (iγ µ ∂µ + M ) from the left gives (4.26) −2 uL uL H (p) 0 or explicitly in the socalled Weyl representation −M E+σ·p E−σ·p −M uR uL = 0. 4. which in general is a linear equation in pµ .31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p).29) which is an explicit realization of the (momentum space) Dirac equation. As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space. p (4.32) suppressing on the RHS the identity matrix in Dirac space.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. γ } ∂µ ∂ν + M 2 ψ(x) = 0. γ ν } = 2 g µν . (4. Since ∂µ ∂ν is symmetric. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). 2 (4. {γ µ . To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. The ﬁrst indeed is solved. = H −1 (p).24) (4. (4.7. this can be rewritten 1 µ ν {γ .

We will see how this is properly implemented (also for bosons!) when quantizing ﬁelds.37) Using the explicit form of γ in the Weyl representation (see Eq. E = +M (twice) and E = −M (twice). . . From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). This can most easily be seen in the particle rest frame. The second problem encountered before. where γ 0 p0 ψ = M ψ 0 −→ Eψ = M γ 0 ψ.The Dirac equation 31 E empty levels M -M filled levels E particle M -M hole Figure 4. 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. γ =ρ ⊗1= γ = iρ ⊗ σ = 0 −1 −σ k 0 2 We (4. (4.34) for the adjoint spinor ψ ≡ ψ † γ0 .g. remains.1: The Dirac sea of negative energy states and antiparticles. but with properties such that it can be annihilated by the particle (e. opposite charge). 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 . Multiplying with γ 0 on the right and pulling it through one ﬁnds a covariant equation ψ(x) iγ µ ∂µ +M ← = 0.3 General representations of γ matrices and Dirac spinors {γ µ . This hole forms an antiparticle with the same mass.36) (4. (4. one sees that there are two positive and two negative eigenvalues. 4. The Dirac sea forms the vacuum. γ ν } = 2 g µν .27).35) is indeed positive and j 0 can serve as a probability density. This problem was solved by Dirac through the introduction of a negative energy sea. (4.39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. The density j 0 = ψγ 0 ψ = ψ † ψ (4. the one of the negative energy states. From the vacuum a particle can be removed. Relying on the Pauli exclusion principle for spin 1/2 particles.

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

ψ= (4. The latter implies that the conjugate of a right-handed spinor. 0 −ǫ 0 0 −ǫ 0 . In S.53) (4. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.R.59) −→ ψ c = C ψ = η ǫ ξ∗ . = 0 ǫ iσ 2 (4.52) η ξ one can apply space inversion. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . C is real and one has C −1 = C † = C T = −C c and [C. is a left-handed spinor. which is again a solution of the Dirac equation. parity.6). ∂ with the solution ψ c (x) = ηc Cψ (x). 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. ∂ (4. Explicitly.g. (C)W. reverse the charge of the particle.R. Some properties of C are C −1 = C † and C T = −C.55) (4. but it does. In any representation a matrix C exist. ∂ (4.57) (4. e. that brings ψ → ψ c . ψR . (C)S.R. for instance. = = 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. ψ= (4. usually to be taken unity.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. called charge conjugation. γ5 ] = 0.) and all representations connected via a real (up to an overall phase) matrix S.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0.g. x = (t. One has ψ c = −ψ T C −1 . T (4. T (4. This symmetry does not change the spin 1/2 nature. (or W.56) (i/ − M ) ψ c (x) = 0.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. x) → x = (t. −iσ 2 0 −ǫ = . such that T Cγµ C −1 = −γµ . Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η .58) where ηc is an arbitrary (unobservable) phase. (i/ − M ) ψ(x) = 0.R. As with parity we look for a transformation.54) e.

s). Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors.The Dirac equation 34 4. s)¯(p. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. ψ(x) = v ±s (p) ei p·x . ¯ ′ ′ Explicit solutions in the standard representation are χs . s)¯(p. 2M 2M −/ + M p v(p. ¯ ¯ † ′ u (p. with E = Ep = + p2 + M 2 . p (4.g. ¯ (4. ⇔ (/ − M ) u(p) = 0. ∂ −iσ · ∇ −i ∂t − M (4. s ) = v (p. ⇔ (/ + M ) v(p) = 0. s) = u(p.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions.63) u(p. where v satisﬁes σ·p −(Ep + M ) v(p) = 0.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. v(p. In an arbitrary frame one has s · p = 0 and s(p) can e. The ¯ ¯ solutions are normalized to u(p. the spin polarization vector in the rest frame is the starting ˆ point. s′ ) = 2M δss′ . s) e−i k·x b(k. s)v(p. 2M 2M (4. s) u = . s). s′ ) = −2M δss′ . (4. s ) = v † (p. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4.62) where χs are two independent (s = ±) two-component spinors. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation).64) (4. s)u(p. the best representation to describe particles at rest is the standard representation. s) = v(p. s′ ) = 2Ep δss′ .1). In that frame is a spacelike unit vector sµ = (0. s)v(p. s) + v(k.60) ψ(x) = 0. s) ei k·x d∗ (k.68) (4. p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle. in which γ 0 is diagonal (see discussion of negative energy states in section 4.61) σ·p −(Ep + M ) There are also two negative energy solutions. s) v = .69) = − s In order to project out spin states.. After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) = s d3 k u(k. be obtained by a Lorentz transformation. (2π)3 2Ek (4. s) = Ep + M σ ·p Ep +M χs . s)v(p. v (p. s)u(p. s) .70) = Ps = ˆ 0 1∓σ·s 2 2 . s) = u(p. s ) = 0. the spinors satisfy C uT (p.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions. ψ(x) = u±s (p) e−i p·x .65) (4. s) and C v T (p. s)u(p. where u satisﬁes −σ · p Ep − M u(p) = 0.

This means that in the solution space for massless fermions the chirality states.74) . λ) = u(p. Explicit examples of spinors are useful to illustrate spin eigenstates. λ = +1/2) = v(p.73) Also for helicity states C uT (p. −p3 −p1 − i p2 1 v(p. +1/2) = √ u(p. λ = + 2 ) = √ E +M − E −M 2 0√ 2 2 √ 0 E +M + E −M √ √ 0√ √ − E +M − E −M 1 1 E+M − E−M 0 √ . while the negative helicity (λ = −1/2) is in essence lefthanded.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . u(p. v(p. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). helicity states. 2 2 0 √ 2 √ 2 − E+M − E−M 0 .48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). (4. +1/2) = √ 0 E +M E+M . One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . v(p. . The chirality projection operators in Eq.71) p − i p2 . λ = −1/2) = √ − E+M √ 0 0 E+M (4. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . etc. −1/2) = √ E + M −(E + M ) 0 . √ 0 u(p. u(p. p ψR/L ≡ PR/L ψ are independent solutions. one has in standard representation: 0 E+M E+M 1 1 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. λ) = v(p. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. λ) and C uT (p. 1 u(p. We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. chirality. For instance with the z-axis as spin quantization axis. λ = − 1 ) = √ √ v(p. 4. γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. Therefore. (4. λ = − 1 ) = √ E + M − E − M . λ = + 1 ) = √ + E +M − E −M . λ). −1/2) = √ (4. Note that for solutions of the massless Dirac equation /ψ = 0. In principle massless fermions can be described by twocomponent spinors. λ = −1/2) = u(p.The Dirac equation 35 (the last equality in the restframe and in standard representation) projects out spin ±1/2 states (check ˆ ˆ this in the restframe for s = z ).

s)(γ ν )kl ul (k ′ .g µρ g νσ + g µσ g νρ ). namely the calculation of the quantity 1 ∗ (¯(k. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ . s′ )γ µ u(k. (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. s′ ).5 γ gymnastics and applications An overview of properties can be found in many text books. we will give one example. more convenient ¯ to use the projection operators introduced earlier. (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A.76) ul (k ′ . (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!. s). s′ )) . Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd.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 ). ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ . k k 2 . s′ )¯i (k ′ .) In order to show the use of the above relations. For this we ﬁrst have to prove (do this) that (¯(k. (10) γ µ γ ν = g µν − i σ µν (11) γ µ γ ν γ ρ = S µνρσ γσ + i ǫµνρσ γσ γ5 with S µνρσ = (g µν g ρσ . (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. k ′ ) = 2 ′ s. s′ )(γ µ )ij uj (k. u u (4. (8) σµν ≡ (i/2)[γµ . a / (3) γ µ γ ρ γ σ γµ = 4 g ρσ or γ µ //γµ = 4 a · b. s)γ ν u(k ′ . s)γ µ u(k ′ . u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ . s′ )) (¯(k. etc. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ). (Use (γ5 )2 = 1 and pull one γ5 through. s′ ) ¯ u s. s)γ µ u(k ′ .The Dirac equation 36 4. In principle one can take a representation and just calculate the quantity u(k.s′ ∗ (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 . s)¯k (k.g µ1 µ3 Tr(γ µ2 γ µ4 · · · γ µn ) + · · · . s)¯k (k. s′ )) = u(k ′ .) (2) Tr(1) = 4. s′ )(γ µ )ij uj (k. (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0.75) Lµν (k. however. s)(γ ν )kl u u s. 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. It is. (6) Tr(γ5 ) = 0. s)γ µ u(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) and v(p. Show that the equality holds.4 Show that in P+ ≡ s=±1/2 u(p.13. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices. s) u p /+M = 2M 2M both sides are projection operators. 4 2 Exercise 4. (4. e. Exercise 4. by letting them act on the four basis spinors 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. Do this e.g. −x) is also a solution of the Dirac equation. 4. where x = (t. 0 if n is odd. s)¯(p. Excercise 4.77) Exercises Excercise 4. x ˜ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. s). x → x. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ .3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. γ σ ] = σ ρσ . H(p) = exp η·σ 2 = M +E+σ·p . of which only the traces of four and two gamma-matrices are nonzero.6 Apply space-inversion.5 Show starting from the relation {γµ . a 4 a · b.g. ˜ Exercise 4. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν .1 Prove Eq. 1 i −i S ρσ = − [γ ρ .

i.7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. The pion has spin 0. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). .5 and M → ∞. me = 0. We will not pursue this chiral freedom at this point and take α = 0. Do you understand why? The calculation of the life-time is completed in chapter 11.8 × 10−6 . Calculate the value of k for M R = 0. ν p Note: how such parametrizations work will be explained in chapter 8. is a stationary solution of the (free) Dirac equation. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e .The Dirac equation 38 Exercise 4. 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 ). namely i / ψ = eiαγ5 ψ. there is no current ﬂowing through the surface of the cavity. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. 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. you need to determine the corresponding conﬁnement condition for ψ. the decay would even be forbidden.66 MeV. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). n It turns out that there is actually a whole family of conditions that provide conﬁnement.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 ). r). (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. Note: To prove this. Excercise 4. For such a cavity nµ = (0. Calculate for now just |M |2 .e.57 MeV. M R = 1. we want to conﬁne this solution to a spherical cavity with radius R.511 MeV). (b) Next. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors. Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139. mµ = 105. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity. In this solution the wave number k = |k| √ = E 2 − M 2 and χm a two-component spinor while j0 and j1 are spherical Bessel functions.

In an arbitrary frame ǫµ (k) with λ = 0.1 Fields for spin 1 In section 3.4) λ=0. ±(1) are obtained by boosting the restframe vectors.7) . µ (2π)3 2E µ (5. 0).6) where Aµ is the four-vector potential (5. λ) . These are solutions of the Klein-Gordon equation with in addition a condition. just as the two-component Dirac spinors did for spin 1/2. A). with ∂µ V µ = 0 (Lorentz condition).3) Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . λ) + ǫ(λ)∗ (k) eik·x c∗ (k.± (λ) d3 k ǫ(λ) (k) e−ik·x c(k. (5.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. −B 1 0 3 With the electromagnetic ﬁeld tensor (5. The vectorﬁeld is therefore (λ) suited to describe spin 1.5) −E B2 . ǫλ ). (2 + M 2 )Vµ = 0.± 5.2) (5. A real vector ﬁeld Vµ (x) then can be expanded in modes as Vµ (x) = λ=0. namely ǫµ (k0 ) = (0. M2 (5. where k = k0 = (M.1) The Lorentz condition implies that k µ ǫµ (k) = 0.2 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. one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . In the restframe. 39 (5. These can describe spin 1 states in the rest-frame. there are then indeed three independent possibilities.Chapter 5 Vector ﬁelds and Maxwell equations 5.

˜ x Note that via parity transformation one obtains another solution. Therefore if k µ = (|k|.e. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. if in Eq. As remarked above. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. The equation in vacuum.8) (5. 5. −i. (5.13) Hence. 0. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x). . transverse or Coulomb gauge). Under a gauge transformation Aµ −→ Aµ + ∂µ χ. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. (5. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. 40 (5.10) the electric and magnetic ﬁelds are unchanged. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ .12) (5. |k|). 0.17) (radiation. AP (x) = Aµ (˜). Taking the divergence of this equiation.16) (5. (5. if ∂µ Aµ = 0 for large times |t|. in which case one has 2Aµ = jµ . j) the Maxwell equations read ∂µ F µν = j ν . (5. For electrodynamics one has furthermore the freedom of gauge transformations. ∂µ Aµ = 0. One possible choice is the gauge A0 = 0 (5.11) This equation is not aﬀected by a gauge transformation.15 the d’Alembertian is replaced by 2 + M 2 . for M = 0. 0). it will vanish at all times. Furthermore. It states the absence of magnetic charges and Faraday’s law.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld.15) (5.5. 2Aµ = 0. only two independent vectors remain ǫ(±) µ (k) = (0. i. (5. in accordance with what we found in section 3.14) of which the solutions give the Li´nard-Wiechert potentials. Although the Lorenz condition is a constraint. ∓1. moree over.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ.

19) are gauge invariant. .0 . In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. This phase diﬀerence is actually observed.Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5.1. but where E = B = 0. may look a bit akward. but A = 0 (hence there is structure in the vacuum). Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. ˆ B = 0. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). The situation. It is however. This occurs in the region outside the solenoid.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. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum). sin ϕ. cos ϕ. 0).1: The Aharonov-Bohm experiment 5. while the electromagnetic ﬁeld Aµ is not. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . (5. λ L − λ λ or ∆x = (L− λ/d)δ. . 0) and ϕ = (− sin ϕ. where Aµ = 0. F = e E + ev × B (5. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. = exp −i The phase diﬀerence between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. nevertheless.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). 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 . illustrated in ﬁg.0 . To be precise. . 2 2 2 BR2 BR2 y BR2 x A= . Inside solenoid: Outside solenoid: B = B z. 5. A= By Bx Br ϕ= − ˆ . This causes an interference pattern as a function of ∆x.

The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 . If this would be be happening in empty space one would be in trouble. Consider a superconductor. the number of times they go around the defect) cannot be continuously deformed into one another. giving no observable phase2 .21) (ϕ being the azimuthal angle). organizing itself in tiny ﬂux tubes (Abrikosov strings). A = ∇χ with χ= BR2 ϕ 2 (5.21. Therefore the gauge transformation must be uniquely deﬁned in the space one is working in.Vector ﬁelds and Maxwell equations U(1) S (space with solenoid) 42 class 0 ei φ class 1 Π1(S) = Z class 2 Figure 5.2: The topology of the space in the Aharonov-Bohm experiment. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. 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. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . 1 The occurrence of nontrivial possibilities. 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. Another situation in which the topology of space can be used is in the case of superconductors. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. ϕ=2π The function χ in Eq. it can be obtained from the situation A = 0 by a gauge transformation. minimizing the space occupied by B ﬁelds. This situation that A can be written as ∇χ is always true when ∇ × A = 0. illustrating also that the group structure of the mapping of U (1) into this space is that of the group of integers Indeed when one considers the A-ﬁeld in the example of the Aharonov-Bohm eﬀect outside the solenoid. Now back to the Aharonov-Bohm experiment.e. Therefore space outside the solenoid doesn’t care that χ is multi-valued. implying that it must be single-valued1 . The AharonovBohm experiment shows a realization of this possibility. In such a space loops with diﬀerent winding number (i. 5.e. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). forming a simple (connected) space. i. however. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually.

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

˙ p= ∂L . and the total change x′ (t′ ) = x(t) + ∆x(t). (6. x (t) = x(t) + δx(t). d dt ∂L ∂x ˙ = ∂L . ∂x (6. x) ˙ (6. ∂x ˙ (6. t1 (6.3) (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. thus t′ = t + δτ.4) δS = t1 t2 dt dt t1 ∂L ∂L δx + δ x + L δτ ˙ ∂x ∂x ˙ d ∂L − ∂x dt ∂L ∂x ˙ δx + t2 t1 = ∂L δx + L δτ ∂x ˙ t2 . As an example. ˙ 2 (6.Chapter 6 Classical lagrangian ﬁeld theory 6.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. These equations can be obtained from a lagrangian using the action principle.8) 44 . recall classical mechanics with the action t2 S= t1 dt L(x. x) = K − V = ˙ 1 mx2 − V (x).5) with ∆x(t) = δx(t) + x(t) δτ . The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6.6) The ﬁrst term leads to the Euler-Lagrange equations.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current.1) and as an example the lagrangian L(x.

t2 ).14) (6. i. λ=0 (6. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. . think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). − d3 x .10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). The resulting variation of the action is δS = R µ ′ = xµ + δxµ . ∂µ φ(x)). 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 . .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . . ∂µ φ(x)) = R d4 x L (φ(x). with ∆φ(x) = δφ(x) + (∂µ φ)δx . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . .6 can be rewritten as δS = . x) ≡ p x − L . which in classical mechanics vanishes at the boundary) does not play a role. R (6. but which in 3 dimensions includes conserved quantities related to rotations and boosts. x). Consider a lagrangian density L which depends on these functions. The hamiltonian H is deﬁned by H(p. . Variations in the action can come from the coordinates or the ﬁelds.12) Here R indicates a space-time volume bounded by (R3 . indicated with δF [φ]/δφ should pose no problems. (R3 . ∂µ φ.9) which is done because the ﬁrst term (multiplying ∆x. x′ ) − d4 x L (φ. ˙ (6. ∂µ φ(x). 6. their derivatives and possibly on the position. ˙ x The second term in Eq. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x).18) 1 Taking a functional derivative. t1 (6. In general any continuous transformation.t1 ) (6. indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). = φ(x) + δφ(x) (6. ∂µ φ′ . δ(∂µ φ) (6. . = ∂R (R3 .11) λ=0 of which p and H are the simplest examples related to space and time translations.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. .2. ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). L (φ(x). dσµ . .t2 ) d3 x . . t1 ) and R3 . + ∂L ∆x − H δτ ∂x ˙ t2 . . We will come back to it in a bit more formal way in section 9.e. H is a conserved quantity.13) (6.Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x.15) d4 x′ L (φ′ . (6. In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds.

the second term is irrelevant. ∂ (6. δL δL ∂µ = . Using the variations in ψ (in the symmetric form). One easily obtains (2 + M 2 )φ(x) = 0.2 Lagrangians for spin 0. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found.25) (6. 2 (6. (2 + M )φ (x) = 0.19) δ(∂µ φ) δφ 6.27) (6.26) The Dirac ﬁeld The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ. 2 ∗ (6. (6. leading to (2 + M 2 )φ(x) = 0. The integrand of the ﬁrst term must vanish.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.29) . L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. (6.23) (6. 2 ψ = 0.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration).24) (6. (6.20) (6. 1/2 and 1 ﬁelds By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar ﬁeld (spin 0). leading to the Euler-Lagrange equations. δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → = ∂ / ψ − M ψ.21) which diﬀer only by surface terms.

(6. 1 M η † ǫη ∗ − M ∗ η T ǫη . ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . . 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 . Peshkin and Schroeder or Exercise 12. 2 (6. (6. Using the twospinors ξ and η. 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.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.31 is given by LM (Dirac) = −M ξ † η + η † ξ . ψL ≡ (ψL )c = C ψL = 0 (6. the mass term in the Dirac lagrangian 6. ∂ ∂ 2 2 (6. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . or equivalently separate the ﬁeld into right.38) (6.g. that the lagrangian separates into two independent parts for M = 0. but note the factor 1/2 as compared to the Dirac lagrangian. and thus (β ∗ α)∗ = α∗ β = −βα∗ . (6. We note that of which the left part coincides . namely2 LM (Majorana) = + (6. In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).36) (6.g.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .32) There exists another possibility to write down a mass term with only one kind of ﬁelds.34) (6.and lefthanded ones.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . it is in the absence c of a Dirac mass term possible to introduce a lagrangian in which only left ﬁelds ψL and ψL appear or one can work with ’real’ spinors or Majorana spinors.31) showing e.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . The reasons for Grassmann variables will become clear in the next chapter. which comes because we in essence use ’real’ spinors.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).7). 2 αβ = −βα. The Majorana case is in fact more general.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0.

Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν .43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x).. which is the space-integral over the zero-component of a conserved current. (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν .47) . (6. These do not aﬀect the dynamics and will give rise to conserved currents..44) δL ∆φ + Θµν (x)δxν . ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. 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. Q= d3 x J 0 (x. (6.46) For quantum ﬁelds (φ operator!) things become more subtle as it it is not possible to specify ˙ for instance φ and φ on the surface σ1 . δ(∂µ φ) (6. Q(t1 ) = Q(t2 ). . t).45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . δ(∂µ φ) (6. requires the presence of a conserved quantity.3 Symmetries and conserved (Noether) currents Next. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e. (6. As an example for the classical ﬁeld.σ1 with σ = (R3 . implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. t). (6. In this case. we consider the second term in Eq. which in a consistent picture precisely generate the symmetries.41) (6. δ(∂µ φ) For continuous transformations. one can consider the variations at the initial or ﬁnal surface. the presence of a symmetry that leaves the lagrangian invariant. we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) . 6. ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). discussed in the next chapter. Returning to δS the surface term is rewritten to δS = R d4 x .40) 6. In particular when the lagrangian is invariant under symmetries.18. these conserved quantities become operators. + ∂R dσµ dσµ ∂R ≡ where d x . ∂µ J µ (x) = 0. .

˙ Q = d3 x j 0 (x. H = d3 x Θ00 (x).53) These are the energy and momentum. t) is the conserved charge (Q = 0). (6. = 0. . Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν .56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . omitting the parameter Λ jµ = − δL δL (i e ψi ) (−i e ψi ) − δ(∂µ ψi ) δ(∂µ ψ i ) 49 = e ψγ µ ψ. ↔ (6.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters.3) j µ = i e φ∗ ∂µ φ. (6. From Noether’s theorem one sees that the current Θµν ǫν is conserved. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds.Classical lagrangian ﬁeld theory From the lagrangian for the Dirac ﬁeld.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. (6. 6. e. we obtain (since δxµ = 0). 1 (6.g.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. The integral over the zero-component. 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. (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . including e the space-time translations and the Lorentz transformations.54) (6.51) (6.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space).4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations.50) (6.49) These currents are conserved as discussed already in chapter 2. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). Summarizing. 2 (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).

69) i 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. one has a tensor that satisﬁes (exercise 6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ .65) M µρσ = T µρ xσ − T µσ xρ . (6. Deﬁning T µν = Θµν − ∂ρ Gρµν .3 we have seen global transformations or gauge transformations of the ﬁrst kind.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 .68) (6.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).58) (6.66) In section 6.59) (6. (6.e.g.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. In general this tensor is not symmetric. however.6) T µν = T νµ . Any lagrangian containing derivatives. • Vector ﬁeld Aµ : −i Sρσ = aρσ . φ(x). ∂µ φ(x) + i e ∂µ Λ(x) e (6. In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. (6. where H ρµν is the quantity appearing in M ρµν .62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. 6. e. The current deduced from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )ij φj + Θµρ xσ − Θµσ xρ 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ . (6. (6. 2 (6. the angle of the transformation depends on the space-time point x. in which the U (1) transformation involves an angle e Λ. ∂µ T µν (6. is not invariant under local gauge transformations.64) (6.3). γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. independent of x. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).70) .67) i.61) A ﬁnal note concernes the symmetry properties of Θµν .63) µν = ∂µ Θ . φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x). −i eΛ(x) ∗ i eΛ(x) (6.

78) . −e j µ Aµ = −e ρφ + e j · A. 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.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).75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld. For this purpose it is necessary to introduce a vector ﬁeld Aµ . The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. ∂µ φ) =⇒ L (φ. (6.73) (6.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. Dµ φ) − Fµν F µν . 2 Minimal substitution ∂µ ψ → (∂µ + i eAµ )ψ leads to the gauge invariant lagrangian L = 1 i ↔ ψ / ψ − M ψψ − e ψγ µ ψ Aµ − Fµν F µν .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 . 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.72) (6. (6. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. As an example consider the Dirac lagrangian. 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). ∂ 2 4 (6. D ∂ A (6. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld. (6.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0. 1 L (φ. 4 (6.76) 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 .

i. φ → eiΛ φ.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . ∂µ F µν = j ν .79) Exercises Exercise 6. give the equation which is satisﬁed by ψ c = Cψ . where j µ = e ψγ µ ψ. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. −eψγ ν ψ. 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.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds.4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. ↔ Exercise 6. ∂µ F µν = j ν . 4 (b) What is the form of the interaction term involving a current jµ and the ﬁeld Aµ that will give the inhomogeneous Maxwell equations. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). (6. Exercise 6.1 (a) Show that the Klein-Gordon equation for the real scalar ﬁeld can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 . giving the Maxwell equation coupling to the electromagnetic current. . and deduce what is the charge of the antiparticle as compared to a particle.e.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. Exercise 6.

.4) for a stationary solution in the non-relativistic (|Enr | ≪ M ) and weak-ﬁeld (|A0 | ≪ M ) limits yields in the standard representation for the ’upper (two) components’ ψu . ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ . Comment: this relation is known as the Pauli equation. 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 . 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. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. This relation is known as the Pauli equation. A) (see also exercise 6. Hint ≡ −gs µe · B = −gs Qe s · B. t) = Enr ψu (r. use ∂µ T µν = 0 and ∂µ M µρσ = 0.4.6 (optional) Prove the properties of the tensor T µν in section 6. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. in order to show that Θρσ − Θσρ = ∂µ H µρσ . 2me Exercise 6. t).Classical lagrangian ﬁeld theory 53 Exercise 6.

starting from the lagrangian L(q. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q. The bilinear product for the conserved quantities.2) H(p. p] = i. Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. (7. q) also considered in the previous chapter. q(q. t). q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. 6.3) For variations δO(p. ˙ 2 (7. is the Poisson bracket [A. namely [q. [q. dλ Examples are dO = [O. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. there are conserved ˙ ˙ quantities Q (Eq. p]P = 1. labeled by the position. H]P dt and dO = [O.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). Quantization of the theory is achieved by imposing canonical commutation relations between q and p. (7. 1 qx (t) = 3 d3 x φ(x. (7. in our example p = mq. q) = p q − L(q. however. Q]P . p]P . Further. dq dp dq dp (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. An immediate generalization for ﬁelds can be obtained by considering them as coordinates.Chapter 7 Quantization of ﬁelds 7.11) found through Noether’s theorem.1 Canonical quantization We will ﬁrst recall the example of classical mechanics for one coordinate q(t). B]P = dB dA dA dB − . q) = ˙ 1 mq 2 − V (q). dq The basic Poisson bracket is actually the one between q and its canonical momentum p.5) ∆ x ∆3 x 54 . ˙ L(q.

Quantization of ﬁelds 55 etc. This implies speciﬁc transformation properties in the Hilbert space for the quantum ﬁelds φ.16) . As an example. ∇φ) (7. (7. ∂µ φ(x)) = ˙ d3 x L (φ. Sometimes it is useful to keep in mind that. ∆φ. py (t)] = i δxy or for the ﬁelds φ(x) and Π(x) the socalled equal time commutation relations [φ(x.12) (7. 2 2 2 (7.7) d3 x L (φ(x). ˙ where Πx (t) is obtained from the continuous ﬁeld Π(x) ≡ δL /δ(∂0 φ). • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). discussed in the section 6. t)] = i δ 3 (x − y). px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t).6) ∆3 x L x (qx . t)] = 0.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum ﬁeld theory. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). i U † (Λ.14) In fact. φ.15) with furthermore the relations [φ(x. Φ(f ) = d4 x φ(x) f (x). 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 . for the scalar ﬁeld theory. the ﬁelds φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f .10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to ﬁnd the canonical momenta.11) (7. albeit with ’sharp’ functions. (7. The essential conditions to consistently quantize a theory then are the following. a) φi (x)U (Λ. qx+∆x ) . ˙ δ φ(x) (7. (7. φ(y.2. t). δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . t).8) (7. t)] = [Π(x. a) = Rj (Λ−1 ) φj (Λ−1 x − a). Π(y. for symmetry transformations.9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . • Poincar´ invariance e The full variation of the ﬁelds. Π(y. formally. t). ∂ qx ˙ ∂ qx ˙ (7. the discretization procedure above is an explicit example.

[N. This must be valid for the – now operators – generators P µ and M µν that in the previous chapter have been found in terms of the ﬁelds through Noether’s theorem. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν .Quantization of ﬁelds 56 or if U (Λ.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† . φ(y)] = 0 if (x − y)2 < 0. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. Q] = [P. [φ(x). [a. a† ] = 0. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . P ] = 0 (7. j (7. a† ] = a† .21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. and [N. Microscopic causality implies local commutativity. i(xµ ∂ν − xν ∂µ )φ (x) + i (7.19) 7. 2 2 (7. Φ(g)] = 0. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ .18) with S µν given in 6.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. a] = −a. (7. Pµ ] = [φ (x). The measurements cannot inﬂuence each other or [Φ(f ). P ] = i.23) The hamiltonian in this case can be expressed in the number operator N = a† a. j 2 one has [φi (x).20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q.4. a] = [a† .17) (Sµν )i φj (x). (7. (7. a† ] = 1. sometimes referred to as second quantization. 2 i ωµν (S µν )i . ω a† a + 2 2 ω i − [Q. P ] 2 (7.25) . [Q. Mµν ] = i i∂µ φi (x).

7. In this way one has found for the harmonic oscillator the spectrum of eigenstates |n (with n a non-negative integer) with En = (n + 1/2)ω. (2π)3 2Ek (7. [φ(x).33) .4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x).e. φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators. is equivalent with [a(k). † √ a and a act as raising and lowering operators. and we see that a state |0 must exist for which N |0 = a|0 = 0.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. which represents the ’number of particles’ with momentum k. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. the real scalar ﬁeld φ(x) becomes d3 k (7.28) (7. N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n .26) a(k) e−i k·x + a† (k) ei k·x . e. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. (7.like’ commutation relations between a(k) and a† (k ′ ).30) [φ(x). but one needs to realize that in that case k 0 = Ek = k2 + M 2 . The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . Note that we will often write a(k) or a† (k).32) (7. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). a† (k ′ )] = 0. φ(x )]x0 =x′0 = [Π(x).31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). i.e. a(k ′ )] = [a† (k). Π(x )]x0 =x′0 = 0. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7. and (7. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . (n − 1) a|n † i.g.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator .

(2π)3 2Ek (7. .35) where the vacuum contribution disappears because of rotational symmetry. (2π)3 (7.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). 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 (7.45) (7. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k). a(k)|0 = 0. i. For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). deﬁning particle states |k = |k (with positive energy.43) The normal ordered product can be expressed as : φ(x)φ(y) := φ+ (x)φ+ (y) + φ− (x)φ+ (y) + φ− (y)φ+ (x) + φ− (x)φ− (y).44) p|φ− (x)|0 = ei p·x .46) (7. is solved by subtracting it as an (inﬁnite) constant. d3 k |k k|.39) The problem with the vacuum or zero-point energy. 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).42) (7. (2π)3 2Ek (7.34) where V = (2π)3 δ 3 (0) is the space-volume. The rest of the states are then obtained from the groundstate via the creation operator. d3 k a(k) e−i k·x . . This term will be adressed below.e. which now contains an inﬁnite number of oscillators. For the momentum operator one has Pi = d3 x Θ0i (x) = 1 2 d3 x ∂ {0 φ ∂ i} φ = d3 k k i N (k). (2π)3 2Ek (7. = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . writing all annihilation operators to the right of the creation operators.41) (7. . in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . k 0 = Ek = k2 + M 2 ). .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. and multiparticle states |(k1 )n1 (k2 )n2 . assuring that the vacuum (by deﬁnition) has eigenvalue 0!. . (2π)3 2Ek d3 k a† (k) ei k·x . |k = a† (k)|0 . (2π)3 2Ek (7. |0 .40) This procedure is known as normal ordering.37) (7. n1 ! n2 ! (7.

e. a† (k ′ )] 3 2E ′ (2π) k [φ− (x). (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. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.56) = −δ 3 (x). d3 k = (2π)3 2Ek = (7. φ(y)] consider µν [φ+ (x). d3 k ei k·(x−y) = −i∆+ (y − x). The last item to be checked for the scalar ﬁeld are the causality condition.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) .49) (7. i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ(y)] which has the properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). (2π)4 (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . x) = 0 and hence ∆(x) = 0 for x2 < 0.Quantization of ﬁelds 59 In order to ensure the consistency of the theory it is necessary to check that the operators P ν and M obtained from the conserved currents Θρσ and M µρσ are indeed the generators of the Poincar´ e group.48) (7.51) (7.50) or as integrals over d4 k. ∆(x) = − ǫ(x0 ) δ(x2 ). 2π (7. In order to calculate [φ(x). (2π)3 2Ek ≡ i∆− (x − y). i. (7.47) ′ d3 k ′ e−i k·x+i k ·y [a(k).52) The result for the invariant commutator function is [φ(x). φ+ (y)] d3 k e−i k·(x−y) . (iii) ∆(0. that they satisfy the required commutation relations in Eqs (7. φ− (y)] ≡ i∆+ (x − y).17) and (7.53) t=0 .18). (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation. = − (2π)3 2Ek (7. (7.

↔ (7. 2 From the lagrangian density (7.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld. (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). (2π)3 2Ek (7. a† (k ′ )] = [b(k). 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) . ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y).63) i.Quantization of ﬁelds 60 7. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x .62) (7. b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). which is equivalent to the relations (only nonzero ones) [a(k). The ﬁeld satisﬁes the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized ﬁelds are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . (2π)3 2Ek (7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). φ† (y)] = i∆(x − y).57) (7. Q = = = i d3 x : φ† ∂0 φ − ∂0 φ† φ(x) : d3 k : a† (k)a(k) − b(k)b† (k) : (2π)3 2Ek d3 k a† (k)a(k) − b† (k)b(k) .e. we will consider it as a repetition of the quantization procedure.65) 7. = ∂{µ φ∗ ∂ν} φ − L gµν .60) = i φ∗ ∂µ φ. (2π)3 2Ek (7.61) (7.66) . extending it with the charge operator and the introduction of particle and antiparticle operators. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero).59) (7.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.

70) where the last line is obtained by using the Dirac equation. s). s)e−i k·x . s)vj (k. The quantized ﬁelds are written ψ(x) = s d3 k b(k. ↔ i ψγµ ∂ν ψ − 2 (7. ∂ 2 ↔ (7.69) (7.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . s)u(k. u v (2π)3 2Ek (7.67) i ψ / ψ − M ψψ gµν . s)ei k·x . s)e−i k·x + d† (k. s) e−i k·(x−y) j s † + vi (k. then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.e. s′ )} = {d(k.74) (7. s)¯(k. s)u† (k. (7. jµ Θµν = = ψγµ ψ. The solution is the introduction of anticommutation relations. (2π)3 2Ek (7. b† (k ′ . ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.77) Note that achieving normal ordering. (7. . † {ψi (x). interchanging creation and annihilation operators.75) (7. d† (k ′ .79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered. s)v(k. s). (2π)3 2Ek d3 k b† (k. s)¯(k. s)ei k·x + d(k.78) (7. {b(k.73) (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ .71) (7. (2π)3 2Ek (7.76) 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.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). i. s) ei k·(x−y) x0 =y 0 .

(7. 4 Π0 Πi δL = 0. ˙ δ A0 δL = F i0 = E i .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. (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. 4 2 This gives the equations of motion discussed before.Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. ∂ (7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum. as the vanishing of Π0 induces a constraint.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before.88) . but has the problem of being noncovariant. For the scalar combination {ψi (x). it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. ˙ δ Ai (7.82) where i∆1 = i∆+ −i∆− . 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). but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. 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). ψ j (y)] = (i/x + M )ij i∆1 (x − y). we would have obtained [ψi (x).86) (7. one has † {ψi (x).84) (7. 7.85) which reﬂects the gauge freedom. ∂ (7. ψ j (y)} at arbitrary times one has {ψi (x).83) From the lagrangian density the canonical momenta are = = (7.6 The electromagnetic ﬁeld 1 L = − Fµν F µν . −λ(∂ρ Aρ ).80) 3 = δ 3 (x − y) δij . It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 .

0.3) satisfy the conditions of a Lie algebra. (7.e. (b) Proof for a quantity O(p. λ)|A = 0. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0.95) Exercises Exercise 7. [ℓi .Quantization of ﬁelds 63 for physical states |A and |B .1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. containing a time-like photon. Q]P . (7. (d) Use Poisson brackets to calculate q and p. which reproduce the Hamilton equations. 0. µ λ=0 (7. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. what are [ri . i. 7. The canonical equal time commutation relations are [Aµ (x).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. (7.e. λ)c(k. pj ]P . λ)ei k·x . The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. for which it is suﬃcient that ∂µ Aµ |A = 0. λ). a longitudinal photon and two transverse photons. 2. q) that dO/dλ = [O. In fact the commutation relations imply ˙ [Aµ (x).11.93) which does exhibit problems with the time-like photon. µ λ=0 (7.91) (7. 0)c(k. ′ (7. 0) . These problems are solved by the Lorentz constraint between physical states given above. and [ℓi . λ)e−i k·x + c† (k. Aν (y)] = [Πµ (x). where Q is the conserved quantity in Eq. and are equivalent with [c(k. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). rj ]P .94) Choosing k µ = (|k 3 |. ˙ ˙ . Πν (y)] = 0.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x).89) with four independent vectors ǫµ . (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. i. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). Aν (y)]x0 =y0 = i gµν δ 3 (x − y). where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. λ) − c† (k. j = 1. It gives 3 k µ ǫ(λ) (k)c(k. pj ]P . 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. c† (k ′ . 6.

(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. (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). ∆hom (Λx) = ∆hom (x) if f is an invariant function. (b) Check the above commutation relation [φ(x).Quantization of ﬁelds 64 Exercise 7. (e) Show that ∆R (x) = 0 if x0 < 0. (2 + M 2 ) ∆hom (x) = 0. . Pµ ] = i∂µ φ(x). (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. (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.3 (a) Show that the correct implimentation of symmetries in the Hilbert space. It is invariant. Is the subtraction of ’zero point’ contributions essential in this check? µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. we can write for ∆R an integral where the k 0 integration remains along the real axis. Give also the expression for ∆C . e−i a requires [φ(x). f (Λk) = f (k) for Lorentz transformations Λ.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.

(h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x). In that case one actually needs discontinuous functions f (k) such as the step function.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). . or ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ). (i) (Optional) Give similar expressions for the other Green’s functions under (d) in terms of the homogeneous solutions under (f). (g) Show that the homogeneous solution ∆ in (f) satisﬁes ∆(0. x) = 0 and argue that one can use Lorentz invariance to show that ∆(x) = 0 for x2 < 0. f (k) ∝ θ(k 0 ).

writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). time reversal (T) and charge conjugation (C).1) The parity operator transforms We will consider the transformation properties for a fermion ﬁeld ψ(x). parity (P). The consequences of P. We can determine A. 8. T and C for classical quantities is shown in the table 1. Both ψ p and ψ satisfy the Dirac equation.1 Parity xµ = (t. (iγ µ ∂µ − M ) ψ(x) = 0.2) 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. x (8. ˜ (8. −r). 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 . T. r) −→ xµ ≡ xµ = (t. Table 8.1: The behavior of classical quantities under P. starting with the Dirac equation for ψ(x).

13) ψ(x) −→ ψ c (x) = ηC C ψ (x).9) (8. −λ)v(k. −λ) e−ik·x − d† (k. x (8. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. ˜ Pop d(k. In the same way as the Fermion ﬁeld. iγ µ† ∂µ − M ψ(˜) = 0. λ)v(k. λ)γ0 v(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0. ˜ −v(k.10) i.5) (8. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. . λ) eik·x . λ)γ0 u(k. −λ)u(k. if helicity λ is used instead of the z-component of the spin m.4) (8. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. λ) Pop = ηP b(k. x The latter behavior of the vector ﬁeld will be discussed further below.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k.e. m). m) γ0 v(k. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). one can also consider the scalar ﬁeld and vector ﬁelds. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). −λ). λ)u(k. m) = = ˜ u(k.6) (check this for the standard representation. (2π)3 2Ek (8. λ) P −1 = −η ∗ d(k. 8. λ) e−ik·˜ + d† (k. λ)u(k. λ)v(k.11) (8.7) = λ d k x x ηP b(k.3) (8. (i˜ µ ∂µ − M ) ψ(˜) = 0.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. op P (8. −λ) e−ik·x − d† (k. −λ). λ) e−ik·x − d† (k. the above operation reverses the sign of λ). m). γ x (8.

λ) eik·x .18) (8.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x). λ). r). λ) e−ik·x + b† (k. (8. λ) −1 Cop = ∗ ηC b(k. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. i. λ) Cop = ηC d(k. it is antilinear1 . m) ¯ C v T (k.21) where A is a 4 × 4 matrix acting in the spinor space. writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . λ) e−ik·x + b† (k. (2π)3 2Ek (8. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. . The (time-reversed) Dirac equation becomes. λ)u(k. x (8. Top |φ = φt | = (|φt )∗ . (8. x (8. to ﬁnd ψ t (−˜) that is a solution of the Dirac equation.3 Time reversal xµ = (t. x iγ µT ∂µ − M ψ(−˜) = 0.16) d k ηC d(k.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k.e. m). γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.14) (8. m) ¯ = = v(k. λ)C uT (k. we start with the complex conjugated x Dirac equation for ψ. λ).Discrete symmetries 68 the latter being also a solution of the Dirac equation. antilinear operator is anti-unitary if A† = A−1 . as discussed in section 4).17) 3 = λ This shows that −1 Cop b(k. λ)C v T (k. The same relations hold for helicity states. 8. ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0.19) Cop d(k. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . x x (8. u(k. r) −→ −˜µ ≡ −xµ = (−t. m). As time reversal will transform ’bra’ into ’ket’. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. λ)v(k.15) (where one must be aware of the choice of spinors made in the expansion. Norm conservation requires Top to be anti-unitary2 .

λ) eik·x + d† (k. λ) T −1 = η ∗ d(k. λ).27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. op T (8. λ)u∗ (k. λ) i γ5 C v(k. p. λ) eik·x + d† (k. λ a |a. The result is = = λ (8. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k.25) (check this for the standard representation). λ)u∗ (k. λ). They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. λ)u∗ (k. (8. λ)v ∗ (k. p. λ) Top = ηT b(k. λ) eik·x + d† (k. ˜ = 0|b(k)|A 8. λ) eik·˜ + d† (k. −p. Since there are 16 independent 4 × 4 matrices.2: The transformation properties of physical states for particles (a) and antiparticles (¯). λ)v ∗ (k. a state |a. λ). there are 16 independent of these bilinear combinations.34) . 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(−˜). −p. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. ∗ ˜ v (k. p.29) In table 2 the behavior of particle states under the various transformations has been summarized.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ). λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ a P |a. −p. λ) e−ik·x (2π)3 2Ek (8.30) (8. −λ a T a. λ)iγ5 Cv(k.31) (8. λ| a. λ |¯. p.32) (8.33) (8. λ)iγ5 Cu(k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ)v ∗ (k. λ| ¯ C |¯. −λ |¯. ˜ Top d(k. −p. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.24) (8.Discrete symmetries 69 Table 8. x (8.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k.26) d k x x ηT b(k.28) (8.

p)u(p) e−i (p−p )·x . σ µν . V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . e. of which the expectation value between momentum eigenstates can be simply found. and T. 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 . γ5 γ µ or γ5 ). F. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. As an example consider the vector current for a point fermion. The 16 combinations of Dirac matrices appearing above are linearly independent. in this case only q 2 = (p′ − p)2 . s′ |V µ (x)|p. C. (q 2 ). This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . called form factors. γ ν ].40) where the coeﬃcients are functions of invariants constructed out of the momenta. (8. p) can be built from any combination of Dirac matrices (1. For instance for nucleons one has Γµ (p′ . In many applications the expectation values of currents are needed.5 Form factors Currents play an important role in ﬁeld theory.. s .Discrete symmetries 70 Table 8.41) 2M 2M . C. F.g. as well as under the combined operation Θ = PCT (see Table 3).35) The x-dependence can be accounted for straightforwardly using translation invariance. (q 2 ). 8..38) (8. γ µ .3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). 2M (8.36) (8. ¯ In general the expectation value between momentum states can be more complicated. (8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ). ¯ ′ ′ (8. for the vector current V µ (x). p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x). p′ . p′ |V µ (0)|p = u(p′ )γ µ u(p).39) where Γµ (p′ .37) (8. p′ |V µ (x)|p = u(p′ )Γµ (p′ . Note that the lagrangian density L (x) → L (−x) under Θ.

the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p) = (i γ5 C)Γ∗ (˜′ . µ p ˜ (8.42) where q = p′ − p. p)γ0 u(p). p) that Γµ (p′ . Therefore hermiticity implies for the structure of Γµ (p′ . p). C and T invariance. ¯ 2M (8. p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p) that γ0 Γµ† (p. p′ )u(p′ ) = u† (p′ )Γµ† (p. or relations based on hermiticity of the operator or P . p)(iγ5 C)u(p). p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ .1 The matrix element of the electromagnetic current between nucleon states is written as < p′ |Jµ (x)|p >= e−i(p−p )·x u(p′ )Γµ u(p) ¯ Here Γµ iσµν q ν F2 (q 2 ) 2M pµ + pµ ′ = γµ H1 (q 2 ) − H2 (q 2 ) 2M = γµ F1 (q 2 ) + ′ . e. p)γ0 . p) that Γµ (p′ .43) = ei q·x u(p′ )γ0 Γµ (˜′ .g.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . or relations that follow from the equations of motion.g.45) Exercises Exercise 8.Discrete symmetries 71 but that are eliminated because of relations between Dirac matrices. p)u(˜) ˜ x p ¯ p p ˜ p (8. p)(i γ5 C). γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . p) = γ0 Γµ (˜′ . p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . p′ )γ0 u(p). p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. e. p′ )γ0 = Γµ (p′ . • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ .

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

1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian.2) (ii) Heisenberg picture. (9. ≡ 0.Chapter 9 Path integrals and quantum mechanics 9.4) ∂ U (t. AH (t) = U −1 (t. Consider the two (time-independent) Heisenberg states: |q. t0 ) ∂t (9. in which the states are time-independent. [AH . t ′ QH (t′ )|q ′ .6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing. t . QH (t)|q. t0 ) = H U (t. t0 ). t0 ) AH (t0 ) U (t. t′ ≡ q ′ |q ′ . t′ .3) (9. H]. (9. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. t ′ |q . i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t0 )|ψS (t0 ) . in which the operators are time-independent. t ≡ q|q. |ψS (t) = U (t. AS (t) = AS and the states o are time dependent.7) 73 . |ψH (t) = |ψH .1 Time evolution as path integral U (t. and the operators are time-dependent. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0.5) (9. (9. (9.

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

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

Dα F [α] = Dα F [α + β]. A useful property of functional integration is the translation invariance. N (9. 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. 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. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. Consider the gaussian functional as an example.28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . Note that procedure (i) is an example of the more general procedure under (ii). 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.25) (see Exercises).26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω . .24) (9. y) = ∗ m.27) (9. 2 (9.Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. α(x) = n αn fn and K(x. with Dα F [α] = n dαn N F (αn ) (9. (9.23) Having deﬁned the Gaussian integral. (9.n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.21) again a multidimensional integral. the following integrals can be derived for a real symmetric or complex hermitean kernel K. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure.

αn = δmn .29) (9.32) (9. which vanish. dx dy θ∗ (x) K(x. θ2 = 0.30) (9.35) For applications to fermion ﬁelds. δα(x) δα(x) δ αKβ = dy K(x.e. αy = δxy . (9. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . n θn fn (x) dx θ∗ (x) θ(x) . however. i.38) . δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . α(y) = δ(x − y). δα(x) (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.31) δ δ α = dy α(y) = dy δ(x − y) = 1. In the expansion of the exponential (from second to third line). y) θ(y) . −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. θη = −ηθ. F [θ. the Gaussian-type functionals. θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1. e. products of the same pair appear. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ.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 ) = 1.33) (9.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ . Examples of functional diﬀerentiation are (9.34) (9. δα(x) or in discretized form δ 1 ∂ ∂ with .g. (9. In principle the deﬁnitions of functionals is the same. (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 = ei Ht |q . T = n J t′ = = Dq exp i t dτ [L(q. L(q. 9. t|Q. the creation of an electron (think of a radio-tube. i. t) φ∗ (Q) e+i En T . for the physical states o q. one has q ′ .Path integrals and quantum mechanics source experiment detector 81 T L t L+Jq t’ L T’ Figure 9. t|n n|Q. Consider. φn (q. say. t|n = q|n e−iEn t with q|n the time-independent wave function. q) + J(τ )q] . furthermore. harmonic oscillator. t |q. t q. We can again rewrite the time-ordered products as functional integrals as derived in the previous section. t′ |q. T . The Heisenberg state |q. q)] . t |T exp −i p exp −i 2π t′ ′ ′ t′ t dτ HI (Q(τ )) |q. t′ |q. ˙ J (9. q) + J(t) · q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). T < t < t′ < T ′ .62) p Dq D exp i 2π t dτ [pq − H(p.64) (9. n (9. T ′ |q ′ . a set |n of physical eigenstates. i. Before and after these processes there is only the vacuum or ground-state |0 . t = = ′ ′ V = q . Considering the source to be present between times t and t′ . T ′ |Q. q . T with q.66) . t is related to the Schr¨dinger state |q by |q. which in turn are embedded between an early time T and a future time T ′ .e. t′ q ′ .1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result. it is possible to switch on an interaction. for example x|n x|n = φn (x) = ei kn x = e −ω 2 x2 /2 (9. ˙ ˙ physically pictured as.s.63) for plane waves. t t′ V =0 Dq D dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t Q′ . q)] ˙ (9. for g. making the electron) and the absorption of an electron (think of a detector).65) J dq ′ dq Q′ .5 The generating functional for time ordered products By introducing a source-term.e. t|Q. T = n q. q) → L(q.

such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞. t|Q. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. thus T →i∞ lim 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 ′ ). eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ . T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . The importance of Z[J] is that the time ordered product of operators can be obtained from it.Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9.70) for the generating functional is in fact ill-deﬁned. t) (9.67) = = φ0 (q. T → i ∞ and T ′ → −i ∞. Since we have (neglecting multiplicative factors). t). t′ |q. J=0 (9.68) (9. = (i)n 0|T Q(t1 ) . 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. Q(tn )|0 . . . T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . . . t′ ) q ′ . δJ(tn ) hence the name generating functional. Better is the use ˜ of imaginary time t = −it.71) 9. (9. ˜ t′ → ∞. 9. . dq dτ + J(τ )q(τ ) .6 Euclidean formulation The above expression (Eq. t 0 0out |0in J . J (9. T = φ0 (q.

ZE [J] = = where LE q. (9.1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. In all of these cases the determinant of the (hermitean) operator K is deﬁned as det K ≡ p Kp . e. δJ(tn ) = (i)n J=0 1 δ n ZE [J] .77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors.74) − V (q) 2 (9. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K.73) d˜ LE q. . ensuring convergence for the functional integral ZE [J]. δJ(tn ) J = 0 ˜ t = it (9. . . dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. As discussed in the previous section the interesting quantities are obtained from functional diﬀerentiation with respect to the sources. when L q.n fm (x) Kmn fn (y). ˜ dt (9. δ n Z[J] 1 Z[J] δJ(t1 ) .72) (9. τ dq ˜ dt ≡ −L q. (9. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q).g. for Grassmann valued functions. det K Exercise 9. complex. Exercises Exercise 9.and Grassmann-valued functions. y) = ∗ m. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . .76) which is a positive deﬁnite quantity. .2 Check the examples of functional derivation for real-.75) = − LE q.Path integrals and quantum mechanics 83 In terms of the imaginary time we can write the Euclidean generating functional. i dq . 1 = √ . ˜ ˜ ZE [J] δJ(t1 ) . + V (q). where Kp are the eigenvalues of the matrix Kmn in K(x. det K 1 . The diﬀerentiations in Z[J] and ZE [J] are related.

y) K −1 (y. z) = δ(x − z).80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . .79) (9.Path integrals and quantum mechanics 84 Exercise 9. det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η . 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 . 2 (9.78) (9.

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 ) . i. . (10.2) = ∞ −∞ d4 xE [LE (φ. x′ . φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . .3) The Euclidean formulation is as before.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.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. . x. .e. t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. Π) (10. ∂µ φ) + Jφ] (10. . .4) J=0 . xn ). .Chapter 10 Feynman diagrams for scattering amplitudes 10. t′ |φ. ∂µ φ) − Jφ] (10. . φ(x. δJ(xn ) 85 ≡ G(n) (x1 . t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ .

. .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). . .9) i. .e. .8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10. (10. xn ) J(x1 ) .10) J=0 In order to determine ∆F . . . As discussed in the previous section (exercise 9. 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 (10.13) . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). with Z0 [0] = 1. . Furthermore. (2π)4 (k 0 + iǫ)2 − E 2 (10. . xn ) = x2 xn . d4 xn G(n) (x1 .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . 2 2 → (10. (10.3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .11) Depending on the path choosen in the complex k 0 plane (see exercise 7.5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . . J(xn ). ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation.12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. .4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (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. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 . These Green functions appear in the expansion Z[J] = n in n! d4 x1 .

14) They are solutions of the inhomogeneous equation.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation. 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. φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). (10. 10.18) Thus we see various solutions. (2π)4 k 2 − M 2 (10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) .20) = exp − 2 where or (see ﬁg.4). (2π)4 (k 0 − iǫ)2 − E 2 (10. µ 1 1 • Firstly.17) d4 k e−i k·x . . the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10. the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. 4 k 2 − M 2 + iǫ (2π) (10. i. t) = 0 for t < 0. in the (well-deﬁned) Euclidean formulation starting with LE = 2 (∂E φ)2 + 2 M 2 φ2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). 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 . (2π)4 k 2 − M 2 (10. ∆F = ∆C .22) (10. t) = 0 for t > 0. 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). (10. φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x).16) C where C is a closed contour in the complex k 0 -plane. the advanced Green’s function ∆A (x) = satisfying ∆A (x.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). ∆(x) = − d4 k e−i k·x . φ+ (0)] = (2π)3 2E are also shown in exercise (7.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.e.15) e−i k·x d4 k .

(4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10.25) x1 x2 x3 = x4 + + .Feynman diagrams for scattering amplitudes 0 1 1 0 0 0 1 1 1 0 1 0 0 0 1 0 1 0 11 00 k 1 0 1 0 k 11 00 E 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 -E 1 0 1 0 1 0 1111111111111111111111111111111111111 0000000000000000000000000000000000000 1111111 1 0000000 0 11111111111111111 00000000000000000 11111111111111 00000000000000 11 00 11 00 1 0 1 0 1 0 11111111 1 00000000 0 1 0 1 0 E 1 0 1 0 1 0 1 0 11 00 1 0 1 0 1 0 1 0 1 0 1 0 0 1 1 0 1 0 1 0 1 0 1 0 88 Figure 10. we have G0 (x1 . we use the fact that i∆F (x − y) = G(2) (x − y) = 0|T φ(x)φ(y)|0 to calculate ∆F via the time-ordered product. φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). . because this is the only quantity entering Z0 [J]. the same result as above. or diagrammatically (4) (10. δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . We ﬁnd (see also section 7. .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). = θ(x0 − y 0 ) 0|[φ+ (x).1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility.24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . x3 . x2 . Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1.

Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar. with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].31) G(2.27) When interactions are present.e.34) Introducing a vertex point to which four propagators are connected. ≡ −i g d4 z (10.32) (10. x2 ) = G0 (x1 .30) and in zeroth order G(0. (10. (2) (0) (10. x2 . Z[J] = = Dφ exp i exp i d4 z LI (10.35) z .g ) (x1 . L (φ) = L0 (φ) + LI (φ). x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. x3 .29) This expression will be the one from which Feynman rules will be derived.g ) (x1 .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. x4 ) = G0 (x1 . x3 . (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. To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . x2 .33) (4) i −i d4 z LI g 4! d4 z 1 δ i δJ(z) 1 exp − i 2 J ∆F J 2 −3 ∆2 (0) + 6i ∆F (0) F + d4 x ∆F (z − x) J(x) 4 d4 x ∆F (z − x) J(x) 1 exp − i 2 J ∆F J . the generating functional can be written. (10. x2 ) = i ∆F (x1 − x2 ) = G(4.g 0 ) = G0 = 1. i.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.

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

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

p′ . out . in = |p1 . . in = p.43) implying as expected as the only nonzero connected Green’s function G(2) (x1 − x2 ) = (−i) c δ 2 W [J] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) = i ∆F (x1 − x2 ). out|α. (10. x4 10. implying the absence of tadpoles. in = p. x2 . in|p′ . e. In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . x4 ) = c 1 x3 z (10. . in = S|α. the vacuum expectation value of the ﬁeld φ(x) is zero. . out = |p′ . out = δαβ ↔ SS † = 1. we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). (2) The one-particle state is invariant (conservation of energy and momentum. (10. out|α. Sβα ≡ β. Proof: 0. (10. . in|S = α. (2) (10. in|SS † |α. ∂µ ∂ µ + M 2 φin (x) = 0. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout .3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. in = |α. considered explicitly. . out| ⇔ |α. in = p.Feynman diagrams for scattering amplitudes 92 i.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. in| (choose ϕ0 = 0). out . . Next. x3 . in|S = 0. thus β. this will be translated to the action on ﬁelds. For the free scalar theory.g ) (x1 . in|S = β. in|S|p′ . translation invariance).47) . out|p′ .44) J=0 For the interacting theory. φout and the interpolating ﬁeld φ(x).e. in = β.g. out = p|p′ . Proof: α. p. out (suppressing except momenta all other 1 m quantum numbers). (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). pn . in into ﬁnal state particle states (out-states) |β. t = −∞ φin (x) t = +∞ φout (x) φ(x) where φin and φout transform under the Poincar´ group as scalar ﬁelds and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . in ⇔ β. one ﬁnds for the connected 4-point Green’s function at order g. then (n) (n) Gsc = Gc for n ≤ 3.45) x2 as the only surviving diagram (see exercises). out| = eiϕ0 0. out|p′ . x1 G(4. out| and S † |α.

out|φout = β.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions. 2 ∂µ ∂ µ + M0 φ(x) = J(x). 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. The time evolution operator (see Eq.g. e.51) (10. a causality condition that can be proven to imply the absence of interactions. this can actually not be used as it would imply [φ(x). The convergence therefore must be weakened to t→−∞ √ (10. The relation between S and Z[J] To establish this relation. as it stands.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. in|φin S → φin S = S φout (x). S is invariant under e Poincar´ transformations: U (Λ. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. the source term Jφ in the lagrangian density is treated in the interaction picture. a)SU −1 (Λ. (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. The relation between S-matrix and in. (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).54) .e. z) φ(z) d4 y d4 z ∆A (x − y) K(y. 9. √ Although the above. a) = S. i.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) .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 ). β. which in a later stage (renormalization) will become more important. (10.52) where Z is a constant.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. z) φ(z). φ(y)] = Z [φin (x). Proof: β. φin (y)] = iZ ∆(x − y). φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) .and out-ﬁelds is: φin (x) = S φout (x)S −1 . (2π)3 2E (10. implies the strong (operator) convergence φ(x) → Zφin (x).

provided [A. we can. which is equal to the expression e in which the creation operators are placed left of annihilation operators. 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. e e ] = [A. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. y) δ δJ(y) : F [J]. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. one has [A. B] is a c-number).61) e φ+ f where the normal ordered expression : e φ f : is used.58) In order to ﬁnd a solution to this equation note that. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. B] and [A.e. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :.57) i [φin (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).59) (10. B]eB . C] are c-numbers. δJ(z) (10. φ(y)] f (y) e e . with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. e = i. S U [J]] = √ Z δ S U [J]. φ(x). [A. B + C]e e . : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10. just as we did for φ(x). δJ(z) (10.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J].60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). express it in terms of advanced and retarded Green’s functions.62) where F [J] is some (arbitrary) functional. z) d4 y d4 z ∆(x − y)K(y. 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− f (10.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. φ(x). Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x). eB ] = [A. B] (for the case that [A. z) δU [J] . t i Ut∞ [J] φ(x)U−∞ [J] (10. z) . (10.63) . (10.

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

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

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

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

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

q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. s4 )γµ u(p.s4 e4 (m) µν (M) L L .77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. s)γν u(k ′ .79) . s1 ) ¯ −igµν 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 ′ .76) 1 . 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 . t2 (10.Feynman diagrams for scattering amplitudes 100 10. s)¯(k. To lowest order (∝ α = e2 /4π) only one diagram contributes. s3 )(−ie)γ µ u(k.s3 . where u ¯ L(m) µν = = 1 2 u(k ′ .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s3 )γ µ u(k.5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. s1 ) u e2 u(p′ . ¯ q2 (10. s′ ) ¯ u s.s′ ′ ′ = 2 kµ kν + 2 kν kµ + q 2 gµν 1 Tr [(/ ′ + m)γµ (/ + m)γν ] k k 2 ′ ′ = 2 kµ kν + kν kµ − gµν (k · k ′ − m2 ) (10.75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current. s′ )γµ u(k.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .s2 . s4 )(−ie)γ ν u(p. s2 ).78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . s2 )|2 ¯ q2 (10.

s3 )γ µ v(p′ . Similarly. s2 (10. s4 )γ ν u(p. .80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s1 ) u(k.s4 = 1 2s × v (k ′ . This is known as crossing symmetry. . (c) Do the same for the connected Green function Gc . s2 ) ¯ ¯ 1 2s u(p. 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. Exercises Exercise 10. s3 ) ¯ ¯ 1 . s2 )γµ u(k.s2 3 . the masses. s1 )γν v(k ′ . s4 ) u e2 v (k ′ . s1 )|2 ¯ s 1 .2.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 .s3 . (4) (4) . To lowest order (∝ α = e2 /4π) only one diagram contributes. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s4 ) v (p′ . . The diagram. for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. 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.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair.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 . (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive.s2 . s2 )γµ u(k. s3 )γ µ v(p′ .

. but the incoming and outgoing momenta are identical. . xn ). . . xn ) ≡ (2π)4 δ 4 (p1 + .Feynman diagrams for scattering amplitudes 102 Exercise 10. . hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . . . . . + pn ). . pn ) only has n − 1 independent variables in it.2 (a) Show that the translation properties of the ﬁelds and the vacuum imply G(n) (x1 + a. 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). .3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. Exercise 10. Show that i the amplitude is symmetric under the interchange of t ↔ u. their sum is zero. Express the contributions in invariants s. xn ) ∝ (2π)4 δ 4 (p1 + . .3 Show that the combinatorial factors found using the rules given in section 10. . . which means overall momentum conservation in Green functions in momentum space. . Note that the function G(n) (p1 .38.4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. . . Pictorially this implies. . pn ). e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. . or if we in general would consider all momenta as incoming. . |M |2 = T11 + T22 − T12 − T21 . . . Exercise 10. . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . t and u. . . . It is of the form −i M = A1 − A2 . in the amplitude (Tij = A∗ Aj ). 1 2 (b) Calculate the quadratic pieces and interference terms. . . xn + a) = G(n) (x1 . . for instance. + pn ) G(n) (p1 .37 and 10.

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

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 √ = . m2 . a (11.12) pb = (mb . a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. m2 .13) (11. = 2 s λ(s.14) One can.16) . 2mb (11.15) Explicit calculation of the reduced two-body phase space element gives dR(s. p1 . 4π s 4π 8π s 4π (11. m2 ). m2 ) a b . Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . m2 . 0). 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. use the ﬁrst relation and the abovementioned threshold value for s to calculate the threshold for a speciﬁc n-particle ﬁnal state in the target rest frame. m2 . ptrf ). m2 .11) (11. m2 ) is a function symmetric in its three arguments. m2 ) = 4(pa · pb )2 − 4p2 p2 . (11. p2 ) = = where λ12 denotes λ(s.8) (11. 4s (11.9) (11. for instance. 2mb λ(s.Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . Also in this case one can express the energy and momentum in the invariants. mb and s). which in the speciﬁc case a b also can be expressed as λ(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 √ = . m2 ) a b . p1 .10) The function λ(s. 2 s s − m2 + m2 a b √ . In that case trf pa = (Ea .

11. pc .25) dR(s. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. there is usually a preferred direction. usually is a 2-particle state. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. pd ) = 4π s 4π 8π s 4π dt dt √ √ .. In inclusive measurements no particles are detected in the ﬁnal state. pT = (px . (11. An often used set of invariants are the Mandelstam variables. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. Writing E = mT cosh y. Examples appear in → π− + p (11. the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. in which one particle is detected. usually many particles are produced. At high energies. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). Instead of the scattering angle.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt.22) (11. 2s (11. a c with q = λab /4s and q ′ = being 0 or 180 degrees. Consider the process a + b → c + d. tmax min λcd /4s. in exclusive measurements all particles are detected. E .21) (11.4).27) with m2 = m2 − p2 = m2 + p2 + p2 . Consider the 1-particle inclusive case. pz = mT sinh y. . py ). 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 .. H1 + H2 → h + X. A speciﬁc reaction channel starts contributing at the threshold value (Eq. The various possibilities are referred to as diﬀerent reaction channels.19) (11. The initial state. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 .18) (11. The variable s is always larger than a c b d the minimal value (ma + mb )2 .Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. however.17) (11. Of course there are not only 2-particle channels. The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd .20) → π0 + n → π+ + π− + n → . = (11.

c bd Mad→c¯(su = u. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. ¯ tt = (pa − p¯)2 = s. b ut = (pa − pd )2 = u. (11. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. t. u).28) which (only involving angles) is easier to determine. 11. 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 = = = . With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . tt = s. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. t. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). ¯ b (11. η = − ln(tan(θ/2)) = tanh−1 (cos θ). This is known as crossing symmetry. t. u). t-channel and u-channel processes respectively. u = (p1 − p4 )2 = (p2 − p3 )2 . ¯ tu = (pa − pc )2 = t. (11.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. which means that rapidity distributions dN/dy maintain their shape. uu = s) = Mab→cd (s.31) . 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.2 Crossing symmetry In the previous chapter. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. tu = s. ut = u) = Mab→cd (s.30) (11. uu = (pa − p¯)2 = s.

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

. . as expected. N 1 .36) (11.e. Using ρtrf = a (0) (0) (0) lab lab lab ρa · γa = ρa · Ea /ma (where ρa is the rest frame density) and va = plab /Ea we have a trf ρa ρb va = (0) ρa ρb 2 λab 4 ma mb (0) (0) or with ρa = 2 ma .34) i. . pn ). dt (11. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. 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 .38) (in which we can calculate −iMf i using Feynman diagrams). (11. it is.37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears. Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . N and T · V are covariant. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ. (11. T · V ρa (11.Scattering theory 108 Although this at ﬁrst sight does not look covariant. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 .39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). σ= √ 2 λab T · V (11. .T V. For the (proper) decay width one thus has 1 N Γ0 = .T d4 x = V · T (2π)4 δ 4 (Pi − Pf ). . (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.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.35) This quantity is not covariant. 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.

p1 . q n ). respectively. (11. pn ).42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s. 2 λab and for instance for two particles dσ = q ′ M (s. . p1 . .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. .Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). . pn ) |M |2 q 32π 2 m2 dΩ |M |2 .e. θ) in quantum mechanics. The result is N = |Mf i |2 dR(s. q n ) q n qn √ Mni (q i . .25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . (11. . ˆ ˆ Inserted in the unitarity condition for M . 16π 2 s (11. . p1 .40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 .47) M (s. q f ). θ). . note the sign and cos θ = qi · qf ).46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. i. for which the √ amplitudes squared have been calculated in the previous chapter. θ) = −8π s ℓ (in analogy with the expansion for f (E. (11. δf n + i(2π)4 δ 4 (Pf − Pn ) (M † )f n or −i Mf i − (M † )f i = − n δni − i(2π)4 δ 4 (Pi − Pn ) Mni = δf i .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. θ) or M (q i . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. (11.43) dΩcm = π M (s. (11. √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). for which one has dσ/dΩ = |f (E. The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude.41) (11. θ)|2 . . . . t) λab 4π dt. 11.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. implies for the scattering matrix M . see 11. i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . pn ). . θcm ) √ q 8π s 2 2 (11. 8π s 2π 8π s .

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

To check unitarity we need the amplitude at order e4 . 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 . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. s − M 2 + iM Γ (11. It is important to note that the physical width of a particle is the imaginary part of the two-point vertex at s = M 2 . 11. The product of amplitudes is of order e4 . . This is what we will do to discuss decay widths. (q Mℓ )ij (s) = −M Γi Γj . . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. however. which is given by a sum of the partial widths into the diﬀerent channels. 0) ∝ e−i(E−iΓ/2)t . For instance the tree-level calculation of Feynman diagrams at order e2 was real. .Scattering theory 111 Note that the total cross section is maximal in the case of full absorption.57) This shows that Γ is the width of the resonance. 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. (11. it is straightforward to write down the partial wave amplitude. σel = σinel . but still real. For an unstable particle one expects that U (t. the time-evolution operator. We note that unitarity is generally broken in a ﬁnite order calculation. The prescription for the pole structure.58) . The quantity Γ is precisely the width for unstable particles. 0).e. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11.5 Unstable particles i k 2 − M 2 + iǫ For a stable particle the propagator is i∆(k) = (11. For the amplitude in a scattering process going through a resonance. i. in which case.55) (note that we have disregarded spin). such that |U (t. .e. ∝ U (t. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . 0)|2 = e−Γt . 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. i. η = 0.56) (again disregarding spin).

Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 .1 Show that the cross section for electron-electron scattering (exercise 10. u) = g(t.88 MeV.59) reaching the unitarity limit for s = M 2 . it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . This is a narrow resonance discovered in 1974. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. which is attributed to neutrinos. • The J/ψ resonance in e+ e− scattering.4) can be written as 4πα2 dσ = {f (t.2 GeV. The cross section at resonance is about 30 nb. This implies that the resonance has spin J = 3/2 (decaying in a P-wave (ℓ = 1) pion-nucleon state) and isospin I = 3/2 (the latter under the assumption that isospin is conserved for the strong interactions). u) + g(t. 2 q (s − M 2 )2 + M 2 Γ2 (11.49 GeV. t) + g(u. u) + f (u. At resonance the cross section σT (π + p) is about 210 mb. This characteristic shape of a resonance is called the Breit-Wigner shape. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. the partial width into e+ e− is Γee = 5.Scattering theory 112 (Prove this using the unitarity condition for partial waves). the full width is Γ = 88 keV. The half-width of the resonance is M Γ. Its mass is M = 1232 MeV. The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ .26 keV. Γ = 2. M2 − s (11. t)} . Exercises Exercise 11.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. its width Γ = 120 MeV. Knowing that each neutrino contributes about 160 MeV (see next chapter). Limiting ourselves to a resonance in one channel. Note that because of the presence of a background the phase shift at resonance may actually be shifted. Three neutrinos explain the resonance shape. • The Z 0 resonance in e+ e− scattering with M = 91. dt s(s − 4m2 ) with f (t. i. From this one sees that a resonance has the same shape in all channels but diﬀerent strength.e. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. where furthermore σinel = 0. All these decays can be seen and leave an ’invisible’ width of 498 MeV. a fast change in the phase shift for a narrow resonance. Its mass is M = 3096.

we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections. As discussed for unstable particles.2 Calculate the ﬂux factor. ρcm ρcm a b 1 cm . Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. writing N (s) M1 (s) = 2 + iM Γ .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.8.999 877 Γ and Γ(π − → e− νe ) = 0.Scattering theory 113 Exercise 11. Exercise 11. Compare the results with the experimental π − lifetime τ = 2. cm |va − vb | in the center of mass system for two colliding particles a and b.2). s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. Calculate now fπ . Exercise 11. Thus. 4.000 123 Γ.4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ .

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

In chapter 10 we have considered quantum electrodynamics as an example of a gauge theory. The photon ﬁeld Aµ was introduced as to render the lagrangian invariant under local gauge transformations. The extension to non-abelian gauge theories is straightforward. The symmetry group is a Lie-group G generated by generators Ta , which satisfy commutation relations [Ta , Tb ] = i cabc Tc , (12.1)

with cabc known as the structure constants of the group. For a compact Lie-group they are antisymmetric in the three indices. In an abelian group the structure constants would be zero (for instance the trivial example of U (1)). Consider a ﬁeld transforming under the group, φ(x) −→ U (θ) φ(x) = ei θ

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

where La is a representation matrix for the representation to which φ belongs, i.e. for a threecomponent ﬁeld (written as φ) under an SO(3) or SU (2) symmetry transformation, φ −→ U (θ) φ = ei θ·L φ = φ − θ × φ.

inf

(12.3)

The complication arises (as in the abelian case) when one considers for a lagrangian density a L (φ, ∂µ φ) the behavior of ∂µ φ under a local gauge transformation, U (θ) = ei θ (x)La , φ(x) ∂µ φ(x) −→ −→ U (θ)φ(x), U (θ)∂µ φ(x) + (∂µ U (θ)) φ(x). (12.4) (12.5)

Introducing as many gauge ﬁelds as there are generators in the group, which are conveniently combined a in the matrix valued ﬁeld W µ = Wµ La , one deﬁnes Dµ φ(x) ≡ ∂µ − ig W µ φ(x), and one obtains after transformation D µ φ(x) −→ U (θ)∂µ φ(x) + (∂µ U (θ)) φ(x) − ig W ′ U (θ)φ(x). µ Requiring that Dµ φ transforms as Dµ φ → U (θ) Dµ φ (or Dµ → U (θ)Dµ U −1 (θ)) gives Dµ φ(x) −→ U (θ)∂µ φ(x) − ig U (θ) W µ φ(x), 114 (12.6)

The standard model

115

**which implies W ′ = U (θ) W µ U −1 (θ) − µ or inﬁnitesimal
**

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

1 1 a ∂µ θa = Wµ + Dµ θa . g g It is necessary to introduce the free lagrangian density for the gauge ﬁelds just like the term −(1/4)Fµν F µν in QED. For abelian ﬁelds Fµν = ∂µ Aν − ∂ν Aµ = (i/g)[Dµ , Dν ] is gauge invariant. In the nonabelian case ∂µ W a − ∂ν W a does not provide a gauge invariant candidate for Gµν = Ga La , ν µ µν as can be checked easily. Generalizing the expression in terms of the covariant derivatives, provides a gauge invariant deﬁnition for Gµν . We have Gµν = with for the explicit ﬁelds transforming like

a a b c Ga = ∂µ Wν − ∂ν Wµ + g cabc Wµ Wν , µν

i [D , D ] = ∂µ W ν − ∂ν W µ − ig [W µ , W ν ], g µ ν

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

1 1 L (φ, ∂µ φ) −→ L (φ, Dµ φ) − Tr Gµν Gµν = L (φ, Dµ φ) − Ga Gµν a 2 4 µν

(12.11)

1 with the standard normalization Tr(La Lb ) = 2 δab . Note that the gauge ﬁelds must be massless, as a 2 a µa mass term ∝ MW Wµ W would break gauge invariance.

**QCD, an example of a nonabelian gauge theory
**

As an example of a nonabelian gauge theory consider quantum chromodynamics (QCD), the theory describing the interactions of the colored quarks. The existence of an extra degree of freedom for each species of quarks is evident for several reasons, e.g. the necessity to have an antisymmetric wave function for the ∆++ particle consisting of three up quarks (each with charge +(2/3)e). With the quarks belonging to the fundamental (three-dimensional) representation of SU (3)C , i.e. having three components in color space ψr ψ = ψg , ψb the wave function of the baryons (such as nucleons and deltas) form a singlet under SU (3)C , 1 |color = √ (|rgb − |grb + |gbr − |bgr + |brg − |rbg ) . 6 (12.12)

The nonabelian gauge theory that is obtained by making the ’free’ quark lagrangian, for one speciﬁc species (ﬂavor) of quarks just the Dirac lagrangian for an elementary fermion, L = i ψ/ ψ − m ψψ, ∂ invariant under local SU (3)C transformations has proven to be a good candidate for the microscopic theory of the strong interactions. The representation matrices for the quarks and antiquarks in the fundamental representation are given by Fa Fa = = λa for quarks, 2 λ∗ − a for antiquarks, 2

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

Figure 12.1: The vectors belonging to internal space located at each point in (one-dimensional) space which satisfy commutation relations [Fa , Fb ] = i fabc Fc in which fabc are the (completely antisymmetric) structure constants of SU (3) and where the matrices λa are the eight Gell-Mann matrices1 . The (locally) gauge invariant lagrangian density is 1 a L = − Fµν F µν a + i ψ/ ψ − m ψψ, D 4 with Dµ ψ = ∂µ ψ − ig Aa Fa ψ, µ (12.13)

a Fµν = ∂µ Aa − ∂ν Aa + g cabc Ab Ac . ν µ µ ν

Note that the term i ψ/ ψ = i ψ/ ψ + g ψ/ a Fa ψ = i ψ/ ψ + j µ a Aa with j µ a = g ψγ µ Fa ψ describes D ∂ A ∂ µ the interactions of the gauge bosons Aa (gluons) with the color current of the quarks (this is again µ precisely the Noether current corresponding to color symmetry transformations). Note furthermore that the lagrangian terms for the gluons contain interaction terms corresponding to vertices with three gluons and four gluons due to the nonabelian character of the theory. For writing down the complete set of Feynman rules it is necessary to account for the gauge symmetry in the quantization procedure. This will lead (depending on the choice of gauge conditions) to the presence of ghost ﬁelds. (For more details see e.g. Ryder, chapter 7.)

**A geometric picture of gauge theories
**

A geometric picture of gauge theories is useful for comparison with general relativity and topological considerations (such as we have seen in the Aharonov-Bohm experiment). Consider the space x x G (called a ﬁbre bundle). At each space-time point x there is considered to be a copy of an internal space G (say spin or isospin). In each of these spaces a reference frame is deﬁned. x ψ(x) denotes a ﬁeld vector ψ(x) which belongs to a representation of G, i.e. forms a vector in the internal space (see ﬁg. 12.1). The superscript x denotes that it is expressed with respect to the frame at point x, i.e. the basis of x G. Let ﬁelds Aa (x) determine the ’parallel displacements’ in the internal space, i.e. connect µ

1 The

Gell-Mann matrices are the eight traceless hermitean matrices generating SU (3) transformations, −i 1 1 i 1 −1 λ3 = λ2 = λ1 = λ4 = 1 1 λ5 = i −i 1 λ6 = 1 1

λ7 =

i

−i

1 1 λ8 = √ 3

−2

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

in reality a not perfectly symmetric situation. we can disregard those ’perturbations’ and look at the ideal situation. this means that there is more than one possibility for the groundstate. Nature will choose one. This characterizes spontaneous symmetry breaking.g. 12. .2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. i. If it is zero one has dxµ Aµ (x) = 0.2. In geometric language the eﬀect on parallel transport of a vector depends on the ’curvature’ Gµν . external magnetic ﬁelds. a theory for a scalar degree of freedom (a scalar ﬁeld) having three (real) components. If the temperature is below a certain critical temperature (T < Tc ) the kinetic energy is no longer dominant and the above hamiltonian prefers a lowest energy conﬁguration in which all spins are parallel. 12.2: The symmetry-breaking ’potential’ in the lagrangian for the case that m2 < 0. As there can be one and only one groundstate. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12. Dν ]. usually being (slightly) prejudiced by impurities. the groundstate itself appears degenerate.The standard model 118 where Gµν = (i/g)[Dµ . Only if this quantity is nonzero a physically observable eﬀect of Aµ exists. φ= 2 φ3 (12. If the temperature is high enough the spins are oriented randomly and the (macroscopic) ground state is spherically symmetric. Nevertheless. e. i>j which is rotationally invariant. In this case there are many possible groundstates (determined by a ﬁxed direction in space).20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj .e. Similarly as the deﬁnition of the covariant derivative the eﬀect is thus frame-independent and we have the transformation law Gµν → S(x)Gµν S −1 (x). which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm eﬀect). 2 2 4 −V (φ) The potential V (φ) is shown in ﬁg. φ1 φ . and equivalently Aµ can be considered as a pure gauge eﬀect.

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. ϕ2 and η. etc. therefore a choice must be made. so one has 0|φquantum |0 = 0 and hence 0|φ|0 = ϕc . The classical groundstate appears degenerate. . The η massless particles are called Goldstone bosons.25) 2 2 2 Therefore there are 2 massless scalar particles. for the vacuum Qa |0 = 0.) terms constitute interaction terms. A quantum ﬁeld theory has only one nondegenerate groundstate |0 .23) F +η the latter only forming a minimum for m2 < 0. Any constant ﬁeld with ’length’ F is a possible groundstate. They are known as the Weyl mode or the Goldstone mode: Weyl mode. (12.e. In this situation one speaks of spontaneous symmetry breaking. (12. The situation now is the following.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the ﬁelds ϕ1 . The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three ﬁelds. while the lagrangian including the nonzero groundstate expectation value chosen by nature.21) Stability of the action requires the classical groundstate ϕc to have a well-deﬁned value (which can be nonzero). λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 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 . say 0 0 . ϕc = 0|φ |0 = (12. The ﬁeld along the third axis plays a special role because of the choice of the vacuum expectation value. In this case the spectrum is described . . In the case of degeneracy. (12. φ = ϕc + φquantum where for the (operator-valued) coeﬃcients in the quantum ﬁeld one wants 0|c† = c|0 = 0. Writing the ﬁeld φ as a sum of a classical and a quantum ﬁeld. The presence of a nonzero value for the classical groundstate value of the ﬁeld will have an eﬀect when the ﬁeld is quantized. It is suﬃcient to do this to second order in the ﬁelds as the higher (cubic. while the quadratic terms must correspond with non-negative masses. i. It is appropriate to redeﬁne the ﬁeld as ϕ1 φ = ϕ2 . In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa .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. The result is 1 1 1 1 (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 − m2 (ϕ2 + ϕ2 ) L = 1 2 2 2 2 2 1 2 1 − m (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12. has less symmetry.

it is easy to see that one gets ∂µ V µ = i ψ [M. . This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. a and a′ are degenerate states. ∂ ∂ (12.e.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. down. i. Neglecting at this point the local color symmetry. etc.26) for all the states labeled by a corresponding to ’broken’ generators.e. . . 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. π (12.. T = τ /2. ψu mu md ψ = ψd . ψ −→ ei α·T ψ. (12. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). A bit more formal. Goldstone mode. [Qa . a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . if the generators Qa generate a symmetry.28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. (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′ . T ] ψ. This means that they are operators that create states from the vacuum. i. The generators Qa (in that case the rotation operators Lz . 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. a massless Goldstone boson for each ’broken’ generator. Using the Dirac equation.The standard model 120 by degenerate representations of the symmetry group. 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. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. irrespective of the fact if they annihilate the vacuum. and one must have mπa = 0. V µ = ψγ µ T ψ. = ψ(i/ − M )ψ. M = . As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. (12. called ﬂavors. . H] = 0. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).31) . strange.e. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). a ∂ µ Jµ = 0. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). i. denoted |π a (k) . Taking the derivative.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices.) one can write L = f ψ f (i/ − Mf )ψf . Including a sum over the diﬀerent ﬂavors (up. there must be a nonzero expectation value 0|Jµ (x)|π (k) .

Note that these transformations also work on the spinor indices. (leaving out the ﬂavor structure) the same as ψγ5 ψ. How can we see this.6 MeV/c2 ). proton and neutron. From group theory (Schur’s theorem) one knows that the latter can only be true.34) Using the Dirac equation. From the number of broken generators it is clear that one expects three massless Goldstone bosons. i. This combined symmetry is what one calls chiral symmetry. Therefore the world of up and down quarks describing pions. T ] = 0. generated by TR/L = 1 (1 ± γ5 )T .e.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. a doublet (isospin 1/2) of nucleons. as the quantity ∂µ Aµ (x). Aµ = ψγ µ T γ5 ψ. the lagrangian density is invariant under left (L) and right (R) rotations independently. parity would not play a role in the world. but the fact that the symmetry is not perfect and the right hand side of Eq. I. parity minus). where the quark masses are not completely zero. while the groundstate is only invariant under isospin rotations (R = L). There exists another set of symmetry transformations. i.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices. is by nature not realized in the Weyl mode. (12. behaves as a pseudoscalar particle (spin zero. chiral symmetry is not perfect. e. for which the ﬁeld (according to the discussion above) has the same behavior under parity. ψ −→ ei α·T γ5 ψ. ψR/L = 1 (1 ± γ5 )ψ.e. socalled axial vector transformations. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . etc. ∂ ∂ (12.27. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . 12.3 MeV/c2 and Mn = 939. This symmetry. which is approximately true for the up and down quarks. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. 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. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . (12. stated equivalently. however. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical.e. with almost degenerate masses (Mp = 938. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. 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. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved.e. if in ﬂavor space M is proportional to the unit matrix. This situation. T γ5 = τ γ5 /2. much smaller than any of the other mesons or baryons. In the real world.e. 12.g. T } γ5 ψ. it is easy to see that one gets ∂µ Aµ = i ψ {M. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. M = m · 1. fπ = 0 remains.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M. both are very small. i. (12. is what happens in the real world. which is equivalent 2 to the V-A symmetry. including in Dirac space a γ5 matrix. etc. but also a triplet (isospin 1) of pions.35 is nonzero. The chiral ﬁelds ψR and ψL are transformed into each other under parity. .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.

.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. (12. The symmetry is broken in the same way as before and the same choice for the vacuum. 0 ϕc = 0|φ |0 = 0 . After symmetry breaking the same number (as expected) comes out. 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ). one sees that the ﬁelds Wµ and Gµν also can be represented as three-component ﬁelds. One may wonder about the degrees of freedom.40) a Dµ φ = ∂µ φ − ig Wµ La φ.The standard model 122 12. . The diﬀerence comes when we reparametrize the ﬁeld φ. F which gives for the lagrangian density up to quadratic terms L 1 1 λ 1 = − Gµν · Gµν + Dµ φ · Dµ φ − m2 φ · φ − (φ · φ)2 4 2 2 4 1 2 2 1 1 2 = − (∂µ Wν − ∂ν Wµ ) · (∂ µ W ν − ∂ ν W µ ) − g F Wµ W µ 1 + Wµ W µ 2 4 2 1 + (∂µ η)2 + m2 η 2 + . (12. in this case three-dimensional) representation reads (La )ij = −i ǫaij . is made. while the number of massive gauge bosons coincides with the number of ’broken’ generators. 4 2 where Since the explicit (adjoint. 0 φ= (12. Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar ﬁeld (Higgs ﬁeld) with three components. but one has 1 massless gauge ﬁeld (2).37) (12. like a photon. 2 massive vector ﬁelds or spin 1 bosons (2 × 3) and one scalar ﬁeld (1). We have the possibility to perform local gauge transformations.39) (12. Therefore we can always rotate the ﬁeld φ into the 3-direction in order to simplify the calculation. made into a gauge theory. having two independent spin components) and three scalar ﬁelds (one degree of freedom each). (12.e. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m . . 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 .42) 2 from which one reads oﬀ that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ). The latter is a spin 0 particle (real scalar ﬁeld) called a Higgs particle. . thus 9 independent degrees of freedom. i. again 9 degrees of freedom.41) φ3 F +η Explicitly one then has 2 2 gF Wµ + g Wµ η 1 1 −gF Wµ − g Wµ η Dµ φ = ∂µ φ + g Wµ × φ = ∂µ η . Dµ φ = Gµν = ∂µ φ + g Wµ × φ. 0 0 0 = . Initially there are 3 massless gauge ﬁelds (each. as in this case there are no massless Goldstone bosons.

43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet. that gives a good description of the physical world. electromagnetic and weak forces. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . τ ). The procedure. 2 (12. down.47) and YW is considered as an operator that generates a U (1)Y symmetry. however. 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.44) which has an SU (2)W symmetry under transformations eiα·TW .45) (12. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature. i. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment.4 The standard model SU (2)W ⊗ U (1)Y The symmetry ideas discussed before play an essential role in the standard model that describes the elementary particles.46) −→ 3 One notes that the charges of the leptons can be obtained as Q = TW − 1/2 for lefthanded particles 3 and Q = TW − 1 for righthanded particles. U(1)Y U(1)Y (12. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. R −→ e−i β 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.) and the gauge bosons responsible for the strong.). (12. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . This is written as 3 Q = TW + YW . 1 As they are massless.50) (12. etc. = ∂µ L − (12. νµ and ντ ) and their antiparticles. neutrinos.51) . while the lefthanded particles form a doublet. 2 2 = ∂µ R + i g ′ Bµ R. the quarks (up. the leptons (elektrons. muons. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L.The standard model 123 12. explicitly L −→ e−i β/2 L. ψR/L = 2 (1 ± γ5 )ψ decouple. etc. its corresponding neutrino (νe . muon µ− or tau τ − ). left. R. (12.48) (12.and righthanded particles. ∂ ∂ ∂ (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 .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.58) (12. † i i g Wµ · τ − g ′ Bµ )φ.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced. leading to the parametrization 1 0 φ(x) = √ (12.55) 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 /2.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . h) (12. + 8 (12. Putting this in leads to L (f ) =L (f 1) +L (f 2) . . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ).54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . (12.59) . . Using local gauge invariance θi for i = 1. . (12. with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. . . this leads to the lagrangian L (h1) . the U (1)Q symmetry (generated by the charge operator). For the classical ﬁeld the choice θ4 = v is made. −V (φ) L and (h2) = −Ge (LφR + Rφ L). TW = ±1 or 0 and YW = 0 (thus Q = TW ) and 3 Bµ is a singlet under SU (2)W (TW = TW = 0) and also has YW = 0. L where L (h1) (h) =L (h1) +L (h2) . (12. which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. 4 4 In order to break the symmetry to the symmetry of the physical world. a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12.

(12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ . 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 .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 .66) From the coupling to the photon. 2 Zµ (12.65) The weak mixing angle is related to the ratio of coupling constants.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .63) (12. . (12. 2 + g ′2 g 3 g ′ Wµ + g Bµ (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. we can read oﬀ e = g sin θW = g ′ cos θW . 4 2 MW g2 v2 = . .69) . 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 + . g ′ /g = tan θW .60) (12. 2θ 2θ 4 cos W cos W 0. 2 (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 .64) (12.The standard model 125 where the quadratically appearing gauge ﬁelds that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ .67) The coupling of electrons or muons to their respective neutrinos.61) (12.

72) (12.5 MeV (exp.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g. With the choosen vacuum expectation value for the Higgs ﬁeld. Finally we want to say something about the weak properties of the quarks.19 GeV.73) (12.74) (12. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) .The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. leptons with CV = −1/2 + 2 sin2 θW ≈ −0. Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three. Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.40 GeV. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). 1. me /v is extremely small.231 2. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.80) First. from the MW mass is about 250 GeV. i. g ′ and v determine a number of experimentally measurable quantities.71) GeV −2 = = = = .79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.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 . as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds).76) with CV CA 3 = TW − 2 sin2 θW Q.166 4 × 10 0.e.78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. v (12.e. (12. g2 MZ 2 2 (CV + CA ). i.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ). such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. (12. . one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh. The coupling to the Higgs particle is weak as the value for v calculated e. 2 2 2 (12. Comparing this with the total width into (invisible!) channels. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. −5 (12. giving rise to a running coupling constant after renormalization of the ﬁeld theory.g. muon or tau. (12. one has the relation g2 e2 1 GF √ = . 80. (12. 91. 3 = TW .77) (12.

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

UR = uR cR tR .g. each of these containing the three families.22 and η ≈ 0. the weak mass eigenstates are L (h2.227. . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . d d (12.85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12.5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2.86) the unitary CKM-matrix introduced above in an ad hoc way.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0. The Higgs ﬁeld is still limited to one complex doublet.83) where Vu and Vd are unitary matrices. Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. e. (12. One then reads oﬀ that starting with the family basis as deﬁned via the left doublets that the mass eigenstates (and states coupling to the Higgs ﬁeld) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . 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. i. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd . The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. 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 .e.82) space. ms and mb ).82 and ρ ≈ 0. u u where we include now the three lefthanded quark doublets in QL .34. φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. (12.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices. † Λu Λ† = Vu G2 Vu . for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . respectively). the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . A ≈ 0.

(12. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . 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. we ﬁnd massless neutrinos. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. (12. e N L= L EL (12. 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 . and obtain L (h2. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . i. As before. since the weak current is the only place where they show up.92) .ℓ) = −LφΛe ER − ER Λ† φ† L. EL = Ve EL .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. The mass ﬁelds ER † mass † mass = We ER . = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. (12.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. there is no family mixing for massless neutrinos.5×10−3 eV2 .ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .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 .e. For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL .90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . 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 . UPMNS a parametrization that in principle also could have been used for quarks.89) with three righthanded neutrinos added to the previous case.The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. In this case. EL = Ve† EL .88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. decoupling from all known interactions. mµ and mτ . one can write Λe = Ve Ge We and we ﬁnd † L (h2. (12. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ).

The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. (12. (12.ν =− containing three (CP-violating) phases (α1 . The mass matrix can be written as ML |MD | eiφ M = (12. 12. 0 0 1 L mass. a unitary matrix relating them to the weak eigenstates.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. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.and righthanded species then just form a Dirac fermion. Thus 2 Υ ′ = nc + nL . one being massive. Actually. Absorption of phases in the states is not possible for Majorana neutrinos. hence the mixing matrix becomes iα /2 0 0 e 1 (12. 1 ∗ c c (12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. For the neutrino sector.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 . as above.99) . α2 and δ). 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. We use here the primes starting with the weak eigenstates. 1 L Υ ′ = nR + nc .and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. however. 1 c ∗ c L mass. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry.94) M L NL NL + M L NL NL . coupled by a Dirac mass term. 2 R ψ = nR + nL . is to add in Eq. are equivalent to two decoupled Majorana neutrinos (see below).The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos.98) (12.97) 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.ν = − MR NR NR + MR NR NR . The way to circumvent this is to introduce as in the previous section righthanded neutrinos. For these light Majorana neutrinos one has. 2 although this option is not attractive as it violates the electroweak symmetry. one must for the neutrinos as well as charged leptons.c. with for the righthanded sector a mass term MR . − MD MR nR 2 (12. For the leptons. This is called the see-saw mechanism (outlined below). couple the right. Thus (disregarding family structure) one has two Majorana neutrinos. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). the left. the massless and massive Majorana neutrinos.95) 2 In order to have more than a completely decoupled sector.

2 with real masses Mi .e.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. related via c Υ1R = U nL . This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc .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)) . T ] ψ. L Υ2L nR Υ1 and Υ2 .105) for i = 1. Υ2R nR (12. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 .e. T } γ5 ψ.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 . H] = 0. i ψ {M. Exercises Exercise 12. (12. 2 U (M M † ) U † = M0 . For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . i.1 Consider the case of the Weyl mode for symmetries. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ . (12. [Qa . Prove that if the generators Qa generate a symmetry. . Exercise 12. The diagonalization is analogous to what was done for the Λ-matrices and one ﬁnds U M U T = M0 with a (unitary) matrix U . a and a′ are degenerate states. 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).2 Derive for the vector and axial vector currents. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † . i.The standard model 131 taking ML and MR real and non-negative. Exercise 12. (12.

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

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

Sign up to vote on this title

UsefulNot usefulrelativistic quantum mechanics

relativistic quantum mechanics

- Quantum Field Theory by Hagen Kleinert
- Brown, Harvey R. & Rom Harre (Eds) - Philosophical Foundations of Quantum Field Theory (1988)
- einstein general relativity
- Quantum Field Theory 2013
- Introduction to quantum field theory, based on the book by Peskin and Schroeder
- An Introduction to Quantum Field Theory - Peskin
- PESKIN - Quantum Field Theory
- Huang K. - Quantum Field Theory
- Srednicki M. Quantum Field Theory - Solutions Manual(Web Draft, 2006)(156s)_PQtb
- An Introduction to Quantum Field Theory(Peskin)
- Srednicki Ms Qft Solutions Rev
- Weinberg, Steven - The Quantum Theory of Fields Volume II
- Lahiri and Pal-A-First Book of Quantum Field Theory-2nd Ed
- Electrical Papers 02 Heav Rich
- Heaviside Papers
- _Review_CH_1-5
- ContinuumMechanics V2
- Non-boolean Computing With Nanomagnetics for Computer Vision Applications
- 978-1-58503-530-4-2.pdf
- Cap_3 VEctores
- BCA(Rev Syll-I-Sem) Assignment for July 2012 and Jan 2013 Sessionn
- [Sloughter_D.]_The_calculus_of_functions_of_severa(Bokos-Z1).pdf
- Force Systems Problems
- Vector Algebra
- Irodov Mechanics
- Lecture 6
- Vectors
- 1. Force Vector
- Weinberg the Quantum Theory of Fields Vol 1 Foundations
- Assembly Design
- QFT-0

Are you sure?

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

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

Get the full title to continue

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