## Are you sure?

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

1980. 2007. Peskin and D. Weinberg. Cambridge University Press.E.V. 3.4 4. Cambridge University Press. Cambridge University Press. 1994.2 36 22 . Addison-Wesly. 4. 1.1 3.1 2 2 3.5 2.2 1.4 2. These notes Section 1.4 3. 5.7 2.3 3.1 6.1 1. McGraw-Hill.H. 3.2 3.4 3.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. Diagrammatica. Peskin & Schroeder and Srednicki. Veltman. 6.3 6. Cambridge University Press. An introduction to Quantum Field Theory.2 3.2 2. 1985. S. but are rather compact.2 6.3 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 2.2 2.3 2. 2. I refer to the book of Srednicki [1] and Ryder [2]. 1995.4 3.5 2. 3. Ryder.1 3. I: Foundations.1 2.1 2.2 3.3 References As most directly related books to these notes.3 3. Vol. The quantum theory of ﬁelds.3.-B.5 5.2 3. Cambridge University Press. Other text books of Quantum Field Theory that are useful are given in refs [3-6]. Itzykson and J. Corresponding chapters in books of Ryder.3 2.3 3.2 4. Srednicki.1 4.2 6.8 3. Quantum Field Theory. M.3 2.2. 2.2 2.2 3. Quantum Field Theory.4 6. Zuber. Quantum Field Theory. C. Schroeder.4 2. II: Modern Applications.5 4.3 4. M.1 5. 1996. The notes contain all essential information.3 3. Vol.1. L.3 3. 1995. M.

6. 2.4 10.3 12.2.3.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.1 5.5 9.4 9.4 2.8 5 4.4 12.1.4 8.1 20.3 3. 2.3 11.3.1 9.4.5 6 7 6 6.4.5 9. 9. 4.1 10.3 20.4 12. 6.2 4.1 12.3.6 3.2 5.5 11.6 3.1 9.5 2.2 9. 6.5 7. 6.2. 4.1 20.4 These notes Section 7.3 8.4 4.2 8. 4.4. 20.1.5.7.3 6.6 10.8 6.5 .5 4.1 7.7 4.6 8.7 5.1.2.3 8.6 3.2.3 7.1 20.2 10.3 9.5 6.2 7. 6.3 7.2 11.1 8.4 7. 9.2 12.2 8.5 4.6 4.3 10.6 6.3 6 8 9 Srednicki 3 3 3 37 5.4 3.5 7.1 11.3 2. 4.6 9. 8.10 8.

5 Corrections 2011/2012) • Extension of Exercise 3.2 . • Slight rewriting of discussion of Majorana mass term in section 6.3 • First two sentences of Chapter 5 have been modiﬁed.

≡ h/2π = 1. In this coordinate representation the position operators r op simply acts as rop ψ(r) = r ψ(r). pj ] = i δij .Chapter 1 Introduction 1. being eigenvalues of the position operators. ˜ ψ(p) = i d3 r exp − p · r ψ(r). c = 299 792 458 m s−1 . One can talk about states |r and the wave function ψ(r) = r||ψ .582 119 15 (56) × 10−22 MeV s. 1 (1. ˜ The corresponding states are |p and the wave functions ψ(p) = p||ψ are the Fourier transforms of the coordinate space wave functions. (1.1 Quantum ﬁeld theory In quantum ﬁeld theory the theories of quantum mechanics and special relativity are united. In quantum mechanics a special role is played by Planck’s constant h. (1. They satisfy the well-known (canonical) operator commutation relations [ri . (1.2) In the limit that v ≪ c one reaches the non-relativistic domain. the position can in principle be determined at any time with any accuracy. (1. In quantum mechanics. the action of the momentum operator in the coordinate representation is not as simple as the position operator. Corresponding position and momentum operators do not commute.4) where δij is the Kronecker δ function. Indeed.3) The uncertainty principle tells us that in this representation the momenta cannot be fully determined.6) . (1.5) One can also choose a representation in which the momenta of the particles are the dynamical variables. It is given by pop ψ(r) = −i ∇ψ(r).1) In the limit that the action S is much larger than . 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. usually given divided by 2π. S ≫ . In special relativity a special role is played by the velocity of light c. quantum eﬀects do not play a role anymore and one is in the classical domain.054 571 68 (18) × 10−34 J s = 6.

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.

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

where m is called the mass of the system. but more important has the same eﬀect on lefthandside and righthandside.1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. deﬁned by ∂2 − ∇2 . ∂1 . (1. = ∂ . The derivative ∂µ is deﬁned ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . . is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . (1. for instance. . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.∇ . a∞ = 4πǫ0 2 /me e2 .28) ν ν Note that gµ with one upper and lower index. ∂t ∂x ∂y ∂z ν ∂µ xν = gµ . g0 = g1 = g2 = g3 = 1. (1. 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.34) (1. (1. ′ ′ .32) .27) It is easy to convince oneself of the nature of the indices in the above equation. explicitly written as (p0 . constructed via the metric tensor itself. = −24. Relate this quantity to the electron Compton wavelength −e via the dimensionless λ ﬁne structure constant α.29) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk .and righthandside a four vector because τ is a scalar quantity! The equation with t = x0 instead of τ simply would not make sense! The momentum four vector. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . p). because one has (1. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. ∂t (1. p) = (E. Note that in this equation one has on left. ∂2 . starting from ǫ0123 = 1.31) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1.36) Exercises Exercise 1. . not aﬀected by Lorentz transformations. 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 .33) (1. 0123 0123 ′ ′ ǫ µνρσ ǫµνρσ The ﬁrst identity. Each of these permutations leads to a minus sign. The length squared of ∂ is the d’Alembertian operator. ∂3 ) = ∂ ∂ ∂ ∂ . It is an invariant tensor. (1.Introduction 6 called covariant.30) (Note that there are diﬀerent conventions around and sometimes the opposite sign is used).35) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ .

n2 . Also construct and calculate the Planck length Lpl . Exercise 1.511 MeV. a2 ] or [a− . 1.3. 2.Introduction 7 (b) Relate the classical radius of the electron (a relativistic concept). (b) The coordinates (a0 . 0) ˆ n2 = (0. a+ . g+− and g−+ . c. a1 . 0) ˆ n0 = (1. ˆ ˆ . 0. σ} is a permutation of {0. which is the Compton wavelength for the Planck mass. n2 . by a simple few-line reasoning [For instance: If {µ. 1. a+ . 1. n3 . One does not need to know . (d) Use the value of the gravitational constant GN / c5 = 6. α and me c2 = 0. Note that ˆ ˆ ˆ ˆ a · nα = aα and derive from that the components nα of the four vector nα . 0. 1) ˆ . ρ.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . This demonstrates how a careful use of units can save a lot of work. n1 . 0. [Note: what is the MKS unit for V/T?] Exercise 1. 1. How are a+ and a− related to a+ and a− .13 to give its actual value in kg. re = e2 /4πǫ0 me c2 to the Compton wavelength. Compare it with the proton mass and use Eq. (c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of c. n+ . a2 . g−− . Use that to give the ˆ ˆµ ˆ components of the basis vectors n+ and n− . e. a3 ) are the expansion coeﬃcients using the basis vectors n0 . ]. 0) ˆ n3 = (0. (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). n1 . Also for the coordinates [a− . 0. a2 ] we can ﬁnd basis vectors n− . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . me . . but only appropriate combinations. a1 . 0.2 Prove the identity A × (B × C) = (A · C) B . Exercise 1. ν.4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . . 3} the index α can only be equal to one of the indices in ǫµνρσ . 0. a1 . a+ . .(A · B) C using the properties of the tensor ǫijk given in section 1. aT ]) are deﬁned through √ a± ≡ (a0 ± a3 )/ 2.71 × 10−39 GeV−2 to construct a mass Mpl (Planck mass). ǫ0 . ˆ ˆ ˆ ˆ These are n1 = (0.

1 and 2. But the replacement 2. t) = ∂2 − ∇2 + M 2 φ(r. t) = exp(−i k 0 t + i k · r). 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. 2M (2.4) where M is the particle mass. one obtains (2.1 The Klein-Gordon equation In this chapter.2.Chapter 2 Relativistic wave equations 2. ∂t 2M (2.2) Acting on the wave function one ﬁnds for a free particle. written as pµ −→ i∂µ is covariant (the same in every frame of reference). Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. t).7) k 0 = ± k2 + M 2 = ±Ek . E = p2 . t) = − ψ(r. Although it is straightforward to ﬁnd the solutions of this equation. ∂t2 (2.5) with (k 0 )2 = k2 + M 2 . φk (r. p2 = pµ pµ = E 2 − p2 = M 2 . (2. p). ∂t p −→ pop = −i∇.3) Equations 2. t) = 0.6) The energy spectrum is not bounded from below: considering the above stationary plane wave solutions. we just want to play a bit with covariant equations and study their behavior under Lorentz transformations.3 are not covariant. 2 + M 2 φ(r.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ . (2. namely plane waves characterized by a wave number k. 8 . i ∂ ∇2 ψ(r. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy.

just as the dependence on time was in ordinary quantum mechanics. At that point the dependence on position r and time t is just a dependence on numbers/parameters on which the operator depends. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.9) They satisfy the continuity equation. it is simply a matter of being careful to ﬁnd a consistent (covariant) theory. −i φ∗ ∇ φ . −k) . (2π)3 2Ek (2. but rather the zero component of a four vector.5). (2. ∂ρ = −∇ · j. Probability is not positive: in quantum mechanics one has the probability and probability current ρ j = = ψ∗ ψ ↔ i i − (ψ ∗ ∇ψ − (∇ψ ∗ )ψ) ≡ − ψ ∗ ∇ ψ. 2.12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. This continuity equation can be written down o covariantly using the components (ρ. 2. however. ↔ ↔ (2. k) ≡ a(k) and 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).3). relativistically the density is not a scalar quantity. for which we need the interpretation of φ itself as an operator. ∂µ j µ = 0.2 Mode expansion of solutions of the KG equation Before quantizing ﬁelds.Relativistic wave equations 9 i. (2. k) + ei k·x φ(−Ek .10) ∂t which follows directly from the Schr¨dinger equation. rather than as a wave function. This leads to the existence of negative densities.8) (2. which is what Lorentz transformations do. It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2.14) ˜ ˜ Introducing φ(Ek . 2M 2M (2. This operator has all possible solutions in it multiplied with creation (and annihilation) operators. there are no longer fundamental objections to mix up space and time. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . we want the most general solution. j) = ↔ i φ∗ ∂0 φ. The KG equation. having the KG equation as a space-time symmetric (classical) equation.13) ˜ with (in principle complex) coeﬃcients φ(k).15) .e.2) is j µ = i φ∗ ∂ µ φ or (ρ. there are solutions with negative energy. Then. As we will see later both problems are related and have to do with the existence of particles and antiparticles. (2. And.11) Therefore. j) of the four-current j. The appropriate current corresponding to the KG equation (see Excercise 2. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2. For this we note that an arbitrary solution for the ﬁeld φ can always be written as a superposition of plane waves. is a second order equation and φ and ∂φ/∂t can be ﬁxed arbitrarily at a given time.

16) The consequence of this is discussed in Exercise 2. The coeﬃcients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. (2π)3 2Ek (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. Performing some Lorentz transformation x → x′ = Λx. −x) ≡ (˜µ ). It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.18) (2. Since a · b = a · ˜ it is easy to see that ˜ b. φP (x) ≡ φ(˜) (P for parity).e. 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) . (2π)3 2Ek (2.17) (2.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x).19) implying that for each solution φ(x) there exists a corresponding solution with the same energy. for which we consider space inversion. i. Another symmetry is found by complex conjugating the KG equation. This shows how parity transforms k-modes into −k modes. x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. It is easy to show that x φP (x) = φ(˜) x = = = or since one can deﬁne φP (x) ≡ d3 k e−i k·x aP (k) + ei k·x bP ∗ (k) . changing the sign of the spatial coordinates.14 and 2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). after restricting the energies to be positive). (2π)3 2Ek (2.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) .6 We will explicitly discuss the example of a discrete symmetry. (2.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. x ∂µ ∂ µ + M 2 φ(˜) = 0. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x). x) → (t. They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). 2. (2. (2.Relativistic wave equations 10 In Eqs 2. The choice of a and b∗ allows an easier distinction between the cases that φ is real (a = b) or complex (a and b are independent amplitudes). which implies (xµ ) = (t.23) .

t) = ei Ek t (i∂0 + Ek ) φ(k. QV ≡ satisﬁes ˙ QV = − ds · j. S V d3 x j 0 (x).Relativistic wave equations 11 i. Use these to show that ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. |f ′ (xn )| . For the real ﬁeld one has aC (k) = a(k). k). Exercise 2.2 Show that if φ and φ∗ are solutions of the KG equation.3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 . d3 k F (Ek . t) for the mode expansion in Eq. 1 Use ↔ ↔ the following property of delta functions δ (f (x)) = zeros xn 1 δ(x − xn ).1 Show that for a conserved current (∂µ j µ = 0) the charge in a ﬁnite volume.e. t). ˙ and thus for any normalized solution the full ‘charge’. t). k) = (2π)4 k2 + M 2 . t) ≡ d3 x φ(x. t) exp(−i k · x) and its time ˜ derivative i∂0 φ(k. (2π)3 2Ek where Ek = Exercise 2. Exercises Exercise 2. ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. letting V → ∞.4 ˜ Write down the 3-dimensional Fourier transform φ(k. ↔ ↔ Exercise 2. This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. is conserved. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k). Q = 0. that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). Note that a(k) and b(k) are independent of t. 2.15.

i (a) Show that relativistic invariance. and β 2 = I. . {αi .e.5 (a) As an introduction to parts (b) and (c). ∂t where ψ is a wave function and the nature of α and β is left open for the moment. B} = AB + BA. an attempt to construct a ﬁrst order diﬀerential equation for the time evolution would be to write.15. B] = AB − BA and the anticommutator as {A.Relativistic wave equations 12 Exercise 2. 2. Is the time dependence of the result as expected? Note that the mode amplitudes will become creation and annihilation operators after quantization of the ﬁelds (Chapter 7). β} = 0. 1 2 (b) Express now the full charge QV (t) (exercise 2. also using the result in (a).6 Write down the mode expansion for the Lorentz transformed scalar ﬁeld φ′ (x) and show that it implies that the Lorentz transformed modes satisfy a′ (Λk) = a(k).1) for a complex scalar ﬁeld current (exercise 2.7 To solve the problem with positive and negative energies and get a positive deﬁnite density. Exercise 2. t) = (−i α · ∇ + m β) ψ(r. (c) Similarly. (b) Show that one has current conservation for the current 2 We denote the commutator as [A. t). From this one concludes that α and β must be matrix-valued and ψ must be a multi-component wave function. One then ﬁnds that Q is expressed as an inﬁnite sum over number operators a† a (particles) and b† b (anti-particles). We will encounter these quantities later as energy and momentum. in terms of the a(k) and b(k). αj } = 2 δ ij I. 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 . a{µ bν} ≡ aµ bν + aν bµ . 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 . ∂ ψ(r. express the quantities E(t) = P i (t) = d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . 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). Exercise 2.2) in terms of the a(k) and b(k) using the expansion in Eq. Note that a{µ bν} indicates symmetrization. j 0 = ψ†ψ and j = ψ † αψ.

however. Since the KG equation expresses just the relativistic relation between energy and momentum. z )V .1) for a rotation around the z-axis or shorthand V ′ = R(ϕ. In quantum mechanics spin is described by a vector. we want to investigate if there exist objects other than just a scalar (real or complex) ﬁeld φ. and we must study the representations of the Lorentz group. in particular each component of these spinors will satisfy this equation.g. Locally this parameter-space is 3-dimensional and correspondingly one has three generators. 0 ≤ ϕ ≤ π. Any rotation can be uniquely written as R(ϕ. For an inﬁnitesimal rotation around the z-axis one has ˆ R(δϕ.3) ϕ=0 13 . in which the parameter space forms locally a Euclidean space.Chapter 3 Groups and their representations The simple systems that we want to describe in a relativistically invariant way are free particles with spin. 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.g. electrons. 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 (3.2) (3. it has 3 components that we know the behavior of under rotations. R(π. e. Particles with spin then will be described by certain spinors. In particular. In a relativistic theory. we also want it to hold for particles with spin. i. The parameter-space of SO(3) is a ˆ ˆ ˆ sphere with radius π. z ) = 1 + i δϕ Lz with as generator ˆ 1 ∂R(ϕ. −n). e. the symmetry group describing rotations is embedded in the Lorentz group.e. The KG equation will actually remain valid. i. 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. z ) Lz = i ∂ϕ 0 = i 0 −i 0 0 0 . 0 0 (3. 3.e.1 The rotation group and SU (2) The rotation groups SO(3) and SU (2) are examples of Lie groups. provided we identify the antipodes. n) where n is a unit vector and ˆ ˆ ϕ is the rotation angle. In this section we will investigate the requirements imposed by Poincar´ e invariance. that is groups characterized by a ﬁnite number of real parameters. n) ≡ R(π.

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

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

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

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

e.24) L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone . which as a matrix equation with (Λ)µν = Λµν and (G)µν = gµν gives (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ .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 (3.26) (3. (ii) |Λ00 | ≥ 1. We considered the generators and looked for representations in ﬁnite dimensional spaces.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. 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 . (3. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.g. it is convenient to use a vector notation for the points in Minkowski space. consisting of the Lorentz group and translations. e To derive some of the properties of the Lorentz group. (det Λ = +1 is called proper. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. Writing x as a column vector and the metric tensor in matrix form. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .25) (3. σ/2 in a two-dimensional (spin 1/2) case.23) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. G = 10 (3. det Λ = −1 is called improper). Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. In this section we consider the Poincar´ group. From this property. 0 0 g00 g01 g02 g03 1 0 g g11 g12 g13 0 −1 0 0 = .27) (3. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.22) (Note that xT is a row vector). 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. proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1.Groups and their representations 18 3.

This is the origin of the Thomas precession. it is easy to ﬁnd the following typical examples from each of the four classes. Returning to the global group. Rotations around the z-axis are given by ΛR (ϕ.31) 0 − 0 +1 0 0 0 0 +1 0 0 +1 0 0 −1 0 0 ∈ L↑ (space inversion) Is = (3.Groups and their representations = i=1 (Λ0 )2 . z ) = exp(−i ηK 3 ). z ) ≈ I + i ϕ J 3 . Thus ˆ 0′ 0 0 0 V 0 V 1 0 0 0 0 1′ 1 V 0 cos ϕ sin ϕ 0 V 0 0 −i 0 2′ = 2 −→ J3 = V 0 − sin ϕ cos ϕ 0 V 0 i 0 0 . 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 . Using these explicit transformations. one distinguishes rotations 0 + and boosts. z ) ≈ ˆ ˆ I − i η K 3 . βγ = sinh η. In L↑ . [K i . which satisfy the commutation relations (check!) [J i . 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. (3. inﬁnitesimally given by ΛB (η. J = (J 1 . J 2 . J 3 ).32) 0 − 0 −1 0 0 0 0 −1 .g. K = (K 1 .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↑ . since the generators K do not form a closed algebra. [J i .28) 3′ 3 V 0 0 0 1 V 0 0 0 0 The parameter η runs from −∞ < η < ∞. z ) = exp(i ϕ J 3 ). i proof: use ΛT GΛ = G and ΛGΛT = G. K j ] = i ǫijk K k . K 3 ). (3. 3 i 2 i=1 (Λ0 ) 3 19 (iii) Boosts along the z-direction are given by ΛB (η. we have found the generators of rotations.30) 0 + 0 +1 0 0 0 0 +1 0 0 −1 0 0 +1 0 0 ∈ L↓ (time inversion) It = (3. J j ] = i ǫijk J k . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. The ﬁrst two sets of commutation relations exhibit the rotational behavior of J and K as vectors in E(3) under rotations. 0 0 +1 0 0 +1 0 0 ∈ L↑ (identity) I = (3. the diﬀerence of ﬁrst performing a boost in the y-direction and thereafter in the x-direction and the boosts in reversed order) contains a rotation (in the example around the z-axis). since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. The commutator of two boosts in diﬀerent directions (e. K j ] = −i ǫijk J k . inﬁnitesimally given by ˆ ΛR (ϕ. K 2 . and those of the boosts.

i i0 i i0 1 ijk M jk 2ǫ (3. Also this group can be divided into four parts. J 3 = M 12 ) and for the (inﬁnitesimal) boosts (along (3.38) Explicitly. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) .34) The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts).g. We therefore ﬁnd (again) that there are six generators for the Lorentz group.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. Of the four parts only L↑ forms a group.39) (e. a). One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . we continue using covariance arguments. This is a + normal subgroup and the factor group is the Vierer group. 3 30 with η = ω and K = M (e. we have for the (inﬁnitesimal) rotations (around z-axis) ΛR = I + i ϕ3 J 3 = I − i ω12 M 12 .36) (3.40) ΛB = I + i η 3 K 3 = I − i ω30 M 30 . Summarizing. K = M ). 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. explicitly reading x′µ = = (Λ)µν xν + aµ = ((I)µν + (ω)µν ) xν + ǫµ Λµν xν + aµ = (g µν + ω µν ) xν + ǫµ . which are obtained by writing the inﬁnitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . inﬁnitesimally approximated by (I + ω. inf (3.37) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation2 .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ.33) These four transformations form the Vierer group (group of Klein). with ϕi = − 1 ǫijk ω jk and J i = 2 z-axis). 3. We will just require Poincar´ invariance for the generators themselves. P+ etc. There are six generators.35) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν . The e 2 These commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . (3. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations. We now want to ﬁnd a covariant form of the six generators of the Lorentz transformations. Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.g. (3. [M µν . ǫ). The extension to the Poincar´ group is e ↑ straightforward. In order to ﬁnd the commutation relations including the translation generators.

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

while the others have no physical signiﬁcance. These e operators are the Casimir operators of the algebra and can be used to deﬁne the representations of ↑ P+ . . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν .52) (3. 0.53) (3. Pν ] = [Wµ . . .56) . Wν ] = −i (gµα Wν − gνα Wµ ) . 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.55) commute with all generators and therefore are invariants under Poincar´ transformations.Groups and their representations 22 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).50) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . 2 exp (−isµ W µ ) |p. (3. . 0. . . the third case represents the vacuum. . From the algebra of the generators one ﬁnds that P 2 = Pµ P µ and W 2 = Wµ W µ (3. p0 > 0. i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. . Massive particles: p2 = M 2 > 0. 2 (3. (3. which can thus be chosen spacelike.51) The following properties follow from the fact that W µ is a four-vector by construction. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . M ǫijk . = = i exp − ǫµνρσ M µν pρ sσ |p. i ǫµνρσ W ρ P σ . Wα ] = [Wµ .49) (3.54) This will enable us to pick a suitable commuting ’spin’ operator. .

1.62) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. s. p0 > 0. (3. i (3. 3.60) and using the completeness relation nµ (p)nν (p) = − g µν − i i 2 i pµ pν M2 . 0). 0. (0. Having made a covariant decomposition. . The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions). 0.58) i. 1 2 (3. 0. .63) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one).e.61) one sees that S i (p) i =− W2 . (1. 2 .e. The commutation relations [W i (p). ms .Groups and their representations 23 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. M J) with components W i = (M/2)ǫijk M jk . . 0).57 The best insight in the meaning of the operators W µ is to sit in the rest frame of the particle and put P µ = (M. Together with the four momentum states thus can be labeled as |M. 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 . S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p).59) (3. 0).64) such that in an arbitrary frame where the four vectors p. M2 (3. |p3 |). 1. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. In that case the vectors ni (p) are just the space directions and W = (0. We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0.. (0. n1 (p). 0. (3. show that W /M can be interpreted as the intrinsic spin. −1) n1 (p0 ) ≡ n2 (p0 ) ≡ s(p0 ) ≡ (3. p. Noting that S i (p) i 2 = 1 W µ (p)W ν (p) M2 niµ (p)niν (p). 0. More generally (fully covariantly) one can proceed by deﬁning S i (p) ≡ W i (p)/M and obtain [S i (p). M2 (3. 0. it is suﬃcient to choose a particular frame to investigate the coeﬃcients in Eq. 1. W j (p)] = i M ǫijk W k (p). n2 (p) and s(p) are obtained by a Lorentz boost from the reference frame one has the property ǫµνρσ sµ nν nρ pσ > 0.57) (i.65) . W i (p) = −W · ni (p)).

P µ . Note that angular momentum does not contribute to helicity as L · p = 0. W 2 (3. This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises). A massless particle.73) (3. We don’t know of any physical relevance. W 2 (p)] = = = i M 2 λ(p). W 1 (p)] [λ(p). i W 2 (p). Thus we have P 2 . however. 1 2 24 (3. thus. W µ (p) = λ(p) P µ .68) (3.70) as a commuting set with zero eigenvalues for P 2 and W 2 . where A = exp(i ϕ · σ/2) = exp(iϕ σ · n/2). implying continuous values for the spin (actually for W i = M S i ). σ · RA a ≡ A(σ · a)A−1 ˆ (valid for any vector a). a)|0 = |0 .66) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). |p| (3. But if M = 0 one has a diﬀerent algebra (Eq. The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. Exercises Exercise 3. −i W 1 (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 . which can take any integer or half-integer value. 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. U (Λ.Groups and their representations The above does actually include the massive case. The algebra of W 1 . Thus 2 λ(p) = J ·p . Wν ]. is characterized by its momentum and the helicity.71) which is called the helicity. W 2 and λ is derived from the general commutation relations for [Wµ . (3.72) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. λ . W 2 (p)] [λ(p).67 has a vanishing right hand side). and gives [W 1 (p). 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).69) We reproduce the algebra for the massive case with W 3 (p) = M λ(p). (3. and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0). This state is in physical applications nondegenerate and is denoted by |0 . |p.67) (3.1 (a) Derive from the relation that deﬁnes the mapping of SU (2) matrices A into SO(3) matrices RA in Eq. 3. The vacuum: pµ = 0. 3.13.

Show that the Lie-algebra representations J and J c are equivalent. Exercise 3. O] + . u). This is true for SU (2) but it is. for instance. one has to explicitly include this rotation. Now write down the three isospin 1 meson quark-antiquark (meson) states. Tr(σi σj σk ) = 2i ǫijk . Both particle and antiparticle have ’spin 1/2’. In both of these cases the symmetry is broken by the charges ¯¯ of the particles. Note the diﬀerence with the (usual) spin 0 and spin 1 states composed from two spin 1/2 states. i. Is this easy to understand.4 (a) If the states in a Hilbert space (unitarily) transform as φ′ = U φ = eiJ φ and O is an operator acting on φ in that Hilbert space. triplet and anti-triplet representation. show that one matrix ǫ exist such that J c = ǫ J ǫ† . what are the matrices J c in the Lie-algebra representation describing the spin rotations for two-component ’anti-spinors’. Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and thus Tr(σi σj ) = 2 δij . d) doublet representation for SU (2)isospin or the Higgs (φ+ . Show actually that ǫ = 2 eiπJ (up to a phase). Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) . show that the transformation for the operators is given by O′ = U O U −1 = O + i [J. φ0 ) doublet representation for SU (2)weak . the trace of the operator Jk is zero (Convince yourself that det(eA ) = eTr A by diagonalizing A. .Groups and their representations 25 for any vectors a and b.e. The appropriate isospin doublet to be used for antiquarks then is (−d. the pions and the isospin 0 meson state. the (non-strange) η-meson state. (b) This establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle. . the operators Jk must be hermitean. Writing A∗ = exp(i α · J c ). and calculate U (a) p U −1 (a) . (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. Examples are the quark (u. (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.2 Show that for a unitary operator U = exp[iαk Jk ] with real coeﬃcients αk . Exercise 3. For a spin 1/2 antiparticle the representation A∗ is needed. (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). not true for SU (3) (ﬂavor or color properties of particles know two distinct deﬁning 3-dimensional representation. Excercise 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. respectively). Show that if det(U ) = 1. .

e. if you don’t want to do this in general. but it can be done. the operator U (u) that gives U (u) r U −1 (u) = r − u t. (c) Give the unitary operator for boosts. = p. i. Exercise 3.6 Consider rotations and translations in a plane (two-dimensional Euclidean space). f (p)] = i ∇p f (p). boost invariance leads to a conserved quantity. . sj ] = iǫijk sk . [ri .69 of the Pauli-Lubanski operators for massless particles. indeed assures d K /dt = 0.e. [ri . Determine the commutation relations for the corresponding operators Ki in quantum mechanics. the others vanish. derive the commutation relations and show that the algebra is isomorphic to (i. H] + i dt ∂O . Do this by showing that the set of operators.5 In classical mechanics (Gallilean invariance) boosts describe invariance of physical laws when viewed from a moving frame. Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. consistent with the (canonical) commutation relations of a quantum theory. momentum and spin operators e satisfy the canonical commutation relations. (a) In classical mechanics. Comment: extending this to more particles is a highly non-trivial procedure. and U (u) p U −1 (u) = p − m u. they describe r −→ r′ = r − u t while t′ = t.e. These do not matter in the non-relativistic limit (c → ∞). show ﬁrst the operator identity [r. = r × p + s. i. although that in this case is a relation that is found already from the Ehrenfest relation for O = ri .67 to 3. From either of these.Groups and their representations 26 Exercise 3. although the presence of an interaction term V (r1 . ∂t Show that boost invariance for a free-moving particle (the commutation relations under (a) being valid). Exercise 3. r2 ) inevitably leads to interaction terms in the boost operators. there is a one to one mapping) the algebra in Eqs 3. sj ] = [pi . For a single free particle or for the free-moving center of mass system that is the quantity K = mr − tp. sj ] = 0. H P J K = p 2 c 2 + m2 c4 . you might just check relations involving J or K by taking some (relevant) explicit components. pj ] = [ri . = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position. rj ] = [pi . Hint: for the Hamiltonian. in particular those with the momentum operator pj and the Hamiltonian H = mc2 + p2 /2m. that’s why many-particle non-relativistic quantum mechanics is ’easy’. pj ] = iδ ij and [si . p×s 1 (rH + Hr) − t p + .7 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations. This is possible for a single particle.. (b) In quantum mechanics one has the Ehrenfest relation i dO = [O.

C): Type I (denoted M ): Type II (denoted M ): 27 J= σ σ .5) = 1 (J + i K).9) (4. K = +i . 2 2 (4.1) (4.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). i ǫijk B k . With this choice of parameter-spaces the plus and minus signs are actually relevant. (4.3) (4. (4. Special cases are the following representations: Type I : (j. B j ] = [Ai . K = i J (A = 0). Aj ] = [B i . 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 . The group SL(2. where we restrict (for ﬁxed n) the parameters 0 ≤ ϕ ≤ 2π and 0 ≤ η < ∞. They will be labeled by two angular momenta corresponding to the A and B groups. C) can be written as a product of a unitary + matrix U and a hermitean matrix H.10) . U (ϕ) H(η) (4. j) K = −i J (B = 0). similarly as the homeomorphism between SO(3) and SU (2). becoming the deﬁning representations of SL(2. 2 1 (J − i K).7) From the considerations above. It is simply connected and forms the covering group of L↑ . j ′ ). (j.6) (4.4) (4. respectively. They precisely correspond to the two types of representations that we have seen before. 0) Type II : (0. K = −i . The matrices in SL(2. C). B j ] = i ǫijk Ak .Chapter 4 The Dirac equation 4. 0. This tells us how to ﬁnd the representations of the group.1 The Lorentz group and SL(2. it also follows directly that the Lorentz group is homeomorphic with the group SL(2. M = exp i ϕ·σ 2 1 exp ± η · σ 2 = U (ϕ) H(η) . C) is the group of complex 2 × 2 matrices with determinant one. 2 (4. 2 2 σ σ J = .8) ˆ ˆ ˆ with η = η n and ϕ = ϕn.

while ±ǫη ∗ transforms as a type-I + spinor. one can use a boost to the frame with momentum ˆ ˆ p.16) = σ µ Λ R aµ . They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that y + ↑ ∗ ǫ ∈ SL(2. U ≡ (U † )−1 = U ) U (ϕ) = exp(i ϕ · σ/2) but diﬀerently under hermitean boosts (H † = H. U σ µ aµ U −1 = σ µ ΛR aµ . (4. This is not true for the two-component spinors in SL(2. ˜ (4. ˜ The sets of four operators deﬁned by deﬁnitions σ µ ≡ (1. thus. ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. however. Λ(ǫ) ∈ L+ . are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . i.13) 2M (E + M ) (exercise 4. (4. Eqs 4.19) .12) ξ → U ξ.18) As already discussed for SU (2).1). cannot be connected by a Lorentz transformation belonging to L↑ . ˜ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. thus. The following relations for + rotations and boosts are useful. C). With E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. This boost then is given by H(p) = exp η·σ 2 = cosh η η ˆ + σ · n sinh 2 2 = M +E+σ·p (4. C) or L↑ . This shows that M ∗ ≃ M . σ).18 show that ±ǫξ ∗ transforms as a type-II (η) spinor.e.17) that can be used to relate σ µ∗ = ǫ† σ µ ǫ = ǫ−1 σ µ ǫ and hence ˜ ¯ U ∗ = ǫ† U ǫ = ǫ† U ǫ.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices. ξ −→ Hξ. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). (4. C). H(η) = exp(−η · σ/2). ǫ−1 = ǫ† = ǫT = −ǫ). ˜ = σ µ Λ B aµ . C) and. η → U η. Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . One deﬁnes spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . such that M = SM S −1 . if there does not exist a unitary matrix S. H ≡ (H † )−1 = H −1 ). (ǫ∗ = ǫ. −σ). (4. σ µ ≡ (1. there exists the matrix ǫ = iσ 2 . The Lorentz transformations Λ(M ) and Λ(M ).15) (4. the ﬁrst equation shows that the conjugate representation with spinors χ∗ transforming according to −σ∗ /2 is equivalent with the ordinary two-dimensional representation (±ǫχ∗ transforms according to σ/2). In fact.11) Considering ξ and η as spin states in the rest-frame. η → Hη. (4.The Dirac equation 28 Let us investigate the deﬁning (two-dimensional) representations of SL(2. U σ µ aµ U ˜ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ µ aµ H ˜ −1 (4. namely H(η) = exp(η · σ/2).

but to L↑ . The behavior is the same for classical quantities. therefore. 0) of SL(2. i.e.17. For a particle at rest only angular momentum is important and we can choose ξ(0. (4. that have the correct commutation relations. where a = (a0 . m) = χm . The i angular momentum operators J are represented by σ i /2. as we have seen in section 4. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). the boost in Eq. −→ −r (vector). 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. (pseudoscalar). i. −→ M (Λ) (4.1. they can be connected.e. For the spin 1/2 representations of the Poincar´ group including parity we. 1 ) and ( 1 . generators. Although these operators cannot be used to label the representations. 0) are inequivalent. 0).The Dirac equation 29 where The matrix I2 does not belong to L↑ . 2 ) and ( 2 . m) = χm uL H(p) 0 . Within the Lorentz group.20) 4.23) u(p. a) and ˜ a = (a0 .22) u= η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq. Thus M ∗ and M are not equivalent within SL(2. must combine the repe resentations. consider the four component spinor that transforms under a Lorentz transformation as 0 ξ ξ . but rather the operators W i (p) should be used. (scalar). 4. This provides the easiest way of seeing ˜ what is happening. C).13. the aij elements of a tensor will not change sign. C) are suitable for representing spin 1/2 particles. but by a transformation belonging ↑ ↑ to the class P− . + − 1 0 0 1 I2 = 0 0 0 0 0 0 −1 0 0 0 0 1 . 1 ) and ( 1 . −a). we obtain for the two components of u which we will refer to as chiral right and chiral left components. we have seen that in the rest frame W i (p)/M → J i . 0 χm uR H(p) . they cannot be connected by a unitary 2 2 transformation. etc. the well-known two-component spinor for a spin 1/2 particle. −→ −p (vector). 4. Taking M (Λ) = H(p). = (4. The representing member of the class P− is the parity or space inversion operator Is . A parity transformation changes aµ into aµ . It has the same eﬀect as lowering the indices. (axial vector).21) In nature parity turns (often) out to be a good quantum number for elementary particle states. e. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. (vector). We also have seen that the representations (0. From the deﬁnition of the representations (0. ) −→ ( . m) = η(0. ǫµνρσ will change sign. 2 2 (4.g.2 Spin 1/2 representations of the Lorentz group 1 1 Both representations (0.

33) deﬁne for a four vector a the contraction a = aµ γµ / . this can be rewritten 1 µ ν {γ . We ﬁrst want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. Since ∂µ ∂ν is symmetric. = H −1 (p). As in section 2 we can use Pµ = i∂µ as a representation for the momenta (translation operators) in function space.25) It is straightforward to eliminate χm and obtain the following constraint on the components of u. This leads to the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space.29) which is an explicit realization of the (momentum space) Dirac equation.32) suppressing on the RHS the identity matrix in Dirac space. The ﬁrst indeed is solved.The Dirac equation 30 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p .28) which is a covariant (linear) ﬁrst order diﬀerential equation. We will discuss another explicit realization of this algebra in the next section. which in general is a linear equation in pµ . (iγ µ ∂µ − M ) ψ(x) = 0. 4. . γ } ∂µ ∂ν + M 2 ψ(x) = 0. (4.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0. (4. p (4. (4. (4. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).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. {γ µ . It is of a form that we also played with in Exercise 2. The explicit realization appearing in Eq.24) (4.27) where γ µ are 4 × 4 matrices called the Dirac matrices1 . = (4.7. 0 H 2 (p) uR uR . ψ † −iγ 0 ∂0 +iγ i ∂i −M 1 We ← ← = 0. γ ν } = 2 g µν .31) To achieve also that the energy-momentum relation p2 = M 2 is satisﬁed for u(p). one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately).27 is known as the Weyl representation. (4. 2 (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0. The general requirements for the γ matrices are thus easily obtained. Applying (iγ µ ∂µ + M ) from the left gives (4. From this one obtains the Cliﬀord algebra for the Dirac matrices.

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

(γµ )S. γ σ ]. For example. −→ Sψ. For instance in the Weyl representation (but valid generally). γ 0 = ρ1 ⊗ 1 = 1 0 γµ ψ Diﬀerent representations can be related to each other by unitary transformations.43) S=√ .47) (4. γ 2 ].46) (4.9 and 4. = S(γµ )W.45) (γ5 )W.41) (4.48) are projection operators.R. (γ5 )S. is 1 1 1 (4. γ µ } = 0 and explicitly one has 0 1 1 . =ρ ⊗1= 0 −1 3 (4.R.49) i 1 with −i S ρσ = 2 σ ρσ = − 4 [γ ρ . that project out chiral right/left states.3) . S −1 . γ 0 ]. K 3 = M 30 = 0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. the rotation and boost generators in Weyl representation in Eqs 4. it is easy to see that PR/L = 1 (1 ± γ5 ) 2 1 0 . 4! (4.R.The Dirac equation 32 The Weyl (or chiral) representation: 0 1 . 0 γ k = −iρ2 ⊗ σ k = k σ −σ k . (4.10 are 1 σ3 0 i J 3 = M 12 = = [γ 1 . 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 . Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.R.40) (4. 2 (4. 0 (4. = ρ ⊗ 1 = 1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . We note that ψ → Lψ and ψ → ψ L−1 . 0 σ3 2 4 1 −iσ 3 0 = − i [γ 3 . −→ Sγµ S −1 .42) We note that the explicit matrix taking us from the Weyl representation to the standard representation. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .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. ν The latter assures Lorentz invariance of the Dirac equation (see Exercise 4. while L−1 γ µ L = Λµ γ ν .

= 0 ǫ iσ 2 (4.g.g. reverse the charge of the particle. is a left-handed spinor. x = (t. 0 −ǫ 0 0 −ǫ 0 .R.58) where ηc is an arbitrary (unobservable) phase. e. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation. (C)S.53) (4. ψR . Explicitly. Some properties of C are C −1 = C † and C T = −C. T (4. One has ψ c = −ψ T C −1 .50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4.52) η ξ one can apply space inversion.54) e.59) −→ ψ c = C ψ = η ǫ ξ∗ . x) → x = (t.R.56) (i/ − M ) ψ c (x) = 0. usually to be taken unity. ψ= (4. This symmetry does not change the spin 1/2 nature. in Weyl representation we ﬁnd in Dirac space T C ξ −ǫ η ∗ . ∂ (4. γ5 ] = 0.R. = = 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. 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. but it does. that brings ψ → ψ c .51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. In any representation a matrix C exist. Note that we have (as expected) explicitly in Weyl x representation in Dirac space P ξ −→ ψ p = γ 0 ψ = η . In S. ∂ with the solution ψ c (x) = ηc Cψ (x). (C)W. C is real and one has C −1 = C † = C T = −C c and [C. (or W.6). such that T Cγµ C −1 = −γµ . called charge conjugation. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. −iσ 2 0 −ǫ = .) and all representations connected via a real (up to an overall phase) matrix S.57) (4.The Dirac equation 33 Parity There are a number of symmetries in the Dirac equation. ψ= (4. The latter implies that the conjugate of a right-handed spinor. which is again a solution of the Dirac equation. As with parity we look for a transformation. ∂ (4.55) (4. T (4. parity. (i/ − M ) ψ(x) = 0. for instance.R.

s) = u(p. s)¯(p. ψ(x) = v ±s (p) ei p·x . s) .69) = − s In order to project out spin states. In an arbitrary frame one has s · p = 0 and s(p) can e. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation). ¯ (4. s ) = v (p.The Dirac equation 34 4. the best representation to describe particles at rest is the standard representation.65) (4. ⇔ (/ + M ) v(p) = 0. (2π)3 2Ek (4. 2M 2M (4. s)u(p. (4.67) It is straightforward to ﬁnd projection operators for the positive and negative energy states P+ P− = s p /+M u(p. s) + v(k. ¯ ¯ † ′ u (p. v (p. where u satisﬁes −σ · p Ep − M u(p) = 0. ∂ −iσ · ∇ −i ∂t − M (4. in which γ 0 is diagonal (see discussion of negative energy states in section 4. ⇔ (/ − M ) u(p) = 0. s) and C v T (p.63) u(p.62) where χs are two independent (s = ±) two-component spinors. s)u(p. s)v(p. s) v = . s). p −σ · p (Ep − M ) Ep + M σ ·p ¯ Ep +M χs χs ¯ For a free massive particle.68) (4. v(p. In that frame is a spacelike unit vector sµ = (0. s)v(p. s) e−i k·x b(k. s). 2M 2M −/ + M p v(p. s)u(p.66) An arbitrary spin 1/2 ﬁeld can be expanded in the independent solutions.64) (4. Note that the spinors in the negative energy modes (antiparticles) could be two diﬀerent spinors. s) u = .61) σ·p −(Ep + M ) There are also two negative energy solutions.60) ψ(x) = 0. ¯ ′ ′ Explicit solutions in the standard representation are χs . with E = Ep = + p2 + M 2 . s ) = 0.70) = Ps = ˆ 0 1∓σ·s 2 2 . s)v(p. s) = u(p. where v satisﬁes σ·p −(Ep + M ) v(p) = 0. s)¯(p.g. s′ ) = 2M δss′ . s) = Ep + M σ ·p Ep +M χs . s) = v(p. It is easy to check that 1 1±σ·s 1 ± γ5 / s ˆ 0 (4. The explicit Dirac equation in the standard representation reads ∂ iσ · ∇ i ∂t − M (4. the spin polarization vector in the rest frame is the starting ˆ point. s′ ) = −2M δss′ . The ¯ ¯ solutions are normalized to u(p. s ) = v † (p. be obtained by a Lorentz transformation. p (4.1). s′ ) = 2Ep δss′ .4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one ﬁnds two solutions. s) ei k·x d∗ (k. After separating positive and negative energy solutions as done in the case of the scalar ﬁeld one has ψ(x) = s d3 k u(k. the spinors satisfy C uT (p. ψ(x) = u±s (p) e−i p·x ..

74) . −1/2) = √ E + M −(E + M ) 0 . p ψR/L ≡ PR/L ψ are independent solutions. λ = + 1 ) = √ + E +M − E −M . This means that in the solution space for massless fermions the chirality states. while the negative helicity (λ = −1/2) is in essence lefthanded. λ = +1/2) = v(p. v(p.71) p − i p2 . λ = − 1 ) = √ √ v(p. u(p. For instance with the z-axis as spin quantization axis. (4. +1/2) = √ 0 E +M E+M . Explicit examples of spinors are useful to illustrate spin eigenstates. The chirality projection operators in Eq. λ) = u(p. p3 E +M E+M 1 2 3 p + ip −p 1 2 p − ip 3 1 −p v(p. One ﬁnds for the helicity states in Weyl representation: √ √ 0√ E +M + E −M √ 1 √ 1 0√ 1 . Therefore. 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 √ . chirality. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). helicity states.72) ˆ Helicity states (p along z ) in Standard representation are: √ √ 0 E+M E +M . −1/2) = √ (4. +1/2) = √ u(p.73) Also for helicity states C uT (p. λ = −1/2) = u(p. etc. . γ5 /ψ = 0 but also p p p /γ5 ψ = −γ5 /ψ = 0. 1 u(p. √ 0 u(p. λ) and C uT (p. v(p. 4. 2 2 0 √ 2 √ 2 − E+M − E−M 0 . λ) = v(p. = √ =⇒ 2E E+M + E−M ( E + M + E − M )2 which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. one has in standard representation: 0 E+M E+M 1 1 0 . We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . u(p. λ = − 1 ) = √ E + M − E − M . −p3 −p1 − i p2 1 v(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 ). λ = −1/2) = √ − E+M √ 0 0 E+M (4. λ). (4.48 replace the spin projection operators which are not deﬁned (by lack of a rest frame). λ = +1/2) = E−M √ 0 − E −M 0 √ √ 0 − E−M − E −M 0 . In principle massless fermions can be described by twocomponent spinors. Note that for solutions of the massless Dirac equation /ψ = 0.

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

by letting them act on the four basis spinors u(p.3 Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν .g. s) u p /+M = 2M 2M both sides are projection operators.2 Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation Exercise 4. γ σ ] = σ ρσ . a 4 a · b. x → x. 4. x ˜ . Show that the equality holds. γν } = 2gµν that γµ / γ µ a = Tr(//) = ab µn µ1 Tr(γ · · · γ ) = −2 /. They lead to Lµν 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] k k 2 = 2 k µ k ′ν + k ν k ′µ − g µν (k · k ′ − m2 ) = = 2 k µ k ′ν + 2 k ν k ′µ + (k − k ′ )2 g µν .4 Show that in P+ ≡ s=±1/2 u(p. i S ρσ ] = g ρµ γ σ − g σµ γ ρ . −x) is also a solution of the Dirac equation. 0 if n is odd. e. s)¯(p. Do this e. 2M (E + M ) ˆ ˆ where E = M γ = M cosh(η) and p = M βγ n = M sinh(η) n. to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜). Excercise 4.6 Apply space-inversion. of which only the traces of four and two gamma-matrices are nonzero. Exercise 4.g.5 Show starting from the relation {γµ . 1 i −i S ρσ = − [γ ρ . by proving that H 2 (p) = exp (η · σ) = (E + σ · p)/M = σ µ pµ /M. where x = (t. ν µ µ Look at the inﬁnitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ . s) and v(p. H(p) = exp η·σ 2 = M +E+σ·p . ˜ Exercise 4.77) Exercises Excercise 4. (4. 4 2 Exercise 4.1 Prove Eq.The Dirac equation 37 The trace is linear and can be split up in parts containing traces with up to four gamma-matrices.13. s).

We will not pursue this chiral freedom at this point and take α = 0. n It turns out that there is actually a whole family of conditions that provide conﬁnement. (b) Next. Note: To prove this.511 MeV). me = 0. Our starting point is the transition amplitude given by M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ). Do you understand why? The calculation of the life-time is completed in chapter 11.5 and M → ∞. |M (π − → µ− νµ )|2 ¯ Actually if the lepton would be massless. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R. ν p Note: how such parametrizations work will be explained in chapter 8.8 × 10−6 . Calculate for now just |M |2 . For such a cavity nµ = (0.66 MeV. n (c) Apply the conﬁnement condition to ﬁnd the value of k for the lowest eigenmode in the spherical ˆ cavity.The Dirac equation 38 Exercise 4. The pion has spin 0. Show that the ratio is |M (π − → e− νe )|2 ¯ ≈ 54. . 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. the lepton and neutrino have spin 1/2 and are ν described by momentum-space spinors.57 MeV. i. r).7 (a) Show that (in standard representation) j0 (kr)χm ψ(x. 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. we want to conﬁne this solution to a spherical cavity with radius R. there is no current ﬂowing through the surface of the cavity. you need to determine the corresponding conﬁnement condition for ψ. Excercise 4. A condition for conﬁning the fermion in the cavity is i / ψ = ψ (where n is the surface normal). mµ = 105. the decay would even be forbidden. Calculate the value of k for M R = 0. namely i / ψ = eiαγ5 ψ. which is the most important factor needed to ﬁnd the decay width (inverse lifetime). is a stationary solution of the (free) Dirac equation. The conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0.e. Show that this n condition is suﬃcient to guarantee that nµ j µ = 0. M R = 1. t) = N k ˆ i E+M σ · r j1 (kr)χm −i Et e .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 ). Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mπ = 139.

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

0. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. (5. j) the Maxwell equations read ∂µ F µν = j ν . 0). while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x).5.11) This equation is not aﬀected by a gauge transformation. moree over.Vector ﬁelds and Maxwell equations and the (conserved) current j µ = (ρ. (5. Although the Lorenz condition is a constraint. The gauge freedom can ﬁrst of all be used to impose the Lorentz condition for the electromagnetic ﬁeld. ∓1. i.10) the electric and magnetic ﬁelds are unchanged. The equation in vacuum. shows that the electromagnetic ﬁelds correspond to the case of a massless spin 1 ﬁeld. |k|). one ﬁnds for a conserved current and λ = 0 2∂µ Aµ = 0. 2Aµ = 0. (5. AP (x) = Aµ (˜). for M = 0. Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector ﬁeld.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. 5. in accordance with what we found in section 3. Under a gauge transformation Aµ −→ Aµ + ∂µ χ. For electrodynamics one has furthermore the freedom of gauge transformations. (5.8) (5. 0. there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). .15) (5. Therefore if k µ = (|k|. It states the absence of magnetic charges and Faraday’s law. (5. it will vanish at all times.12) (5. transverse or Coulomb gauge). ∂µ Aµ = 0.18) corresponding to the helicity states λ = ±1 of the massless photon ﬁeld.14) of which the solutions give the Li´nard-Wiechert potentials. if ∂µ Aµ = 0 for large times |t|. if in Eq.16) (5. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. only two independent vectors remain ǫ(±) µ (k) = (0. −i. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν .e. ˜ x Note that via parity transformation one obtains another solution. Furthermore. One possible choice is the gauge A0 = 0 (5. As remarked above. 40 (5. Taking the divergence of this equiation.17) (radiation. The equations of motion for the ﬁelds Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ .13) Hence.15 the d’Alembertian is replaced by 2 + M 2 . The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . in which case one has 2Aµ = jµ .

.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. ϕ= − ˆ 2r 2r2 2r2 where r = (cos ϕ. 0).0 . Inside solenoid: Outside solenoid: B = B z. Nevertheless the signiﬁcance of Aµ is shown in the Aharonov-Bohm experiment. A= By Bx Br ϕ= − ˆ .0 . F = e E + ev × B (5. illustrated in ﬁg. nevertheless.Vector ﬁelds and Maxwell equations 41 L ∆x d a (infinitely) long solenoid Figure 5. 0) and ϕ = (− sin ϕ. This occurs in the region outside the solenoid. sin ϕ. In the presence of an electromagnetic ﬁeld Aµ an additional phase diﬀerence is observed. In this experiment an observable phase diﬀerence is measured for electrons moving through ﬁeld-free space (E = B = 0 but Aµ = 0). This causes an interference pattern as a function of ∆x.20) e (∇ × A) · ds = B · ds = where Φ is the ﬂux through the solenoid. It is however. The explanation of the signiﬁcance of the vector ˆ ˆ potential A for the phase of the electron wave function is found in minimal substitution ψ ∝ exp i p·r −→ exp i (p − eA) · r e A · r exp i p·r . while the electromagnetic ﬁeld Aµ is not. λ L − λ λ or ∆x = (L− λ/d)δ. The phase diﬀerence between the electrons travelling two diﬀerent paths is given by δ = 2π a a ∆x d = −≈ . (5. . nothing more than a manisfes1 tation of a nontrivial vacuum (a la the Dirac sea for fermions). may look a bit akward. = exp −i The phase diﬀerence between the two paths of the electron then are given by ∆δ = − = e 1 A · dr + e 2 A · dr = e e A · dr e Φ. This phase diﬀerence is actually observed. but A = 0 (hence there is structure in the vacuum). 5. 2 2 2 BR2 BR2 y BR2 x A= . . where Aµ = 0. cos ϕ. The situation. ˆ B = 0.1.1: The Aharonov-Bohm experiment 5. To be precise. but where E = B = 0. The energy density 2 (E 2 + B 2 ) = 0 (hence a vacuum).19) are gauge invariant.

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. organizing itself in tiny ﬂux tubes (Abrikosov strings). Therefore the gauge transformation must be uniquely deﬁned in the space one is working in. Below the critical temperature the magnetic ﬁeld is squeezed out of the superconducting material. The diﬀerence here is that we are working in a space with a ’defect’ (the inﬁnitely long solenoid). A = ∇χ with χ= BR2 ϕ 2 (5.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. it can be obtained from the situation A = 0 by a gauge transformation. is multivalued as ϕ = 0 and ϕ = 2π are the same point in space. In such a space loops with diﬀerent winding number (i. the number of times they go around the defect) cannot be continuously deformed into one another. The AharonovBohm experiment shows a realization of this possibility. This situation that A can be written as ∇χ is always true when ∇ × A = 0. 5. The observable phase going around in general is ∆δ = e A · dr = e ∇χ · dr = e χ|ϕ=0 .e. forming a simple (connected) space. implying that it must be single-valued1 . 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. 1 The occurrence of nontrivial possibilities. 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. i. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . minimizing the space occupied by B ﬁelds. nonobservable phases ϕ = 2π n. however. If this would be be happening in empty space one would be in trouble. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. Therefore space outside the solenoid doesn’t care that χ is multi-valued. Consider a superconductor.e. Another situation in which the topology of space can be used is in the case of superconductors. ϕ=2π The function χ in Eq.21.21) (ϕ being the azimuthal angle). Now back to the Aharonov-Bohm experiment.2: The topology of the space in the Aharonov-Bohm experiment. giving no observable phase2 .

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

(6.Chapter 6 Classical lagrangian ﬁeld theory 6. recall classical mechanics with the action t2 S= t1 dt L(x.5) with ∆x(t) = δx(t) + x(t) δτ .7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. ˙ 2 (6. d dt ∂L ∂x ˙ = ∂L . x) ˙ (6. These equations can be obtained from a lagrangian using the action principle.3) (6. The requirement δS = 0 with ﬁxed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 ′ (6. As an example. Dirac ﬁelds (Dirac equation) and massless vector ﬁelds (Maxwell equations) and corresponding to these ﬁelds conserved currents describing the probability and probability current. x) = K − V = ˙ 1 mx2 − V (x). x (t) = x(t) + δx(t).1) and as an example the lagrangian L(x.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 . ∂x (6.8) 44 . and the total change x′ (t′ ) = x(t) + ∆x(t). thus t′ = t + δτ.6) The ﬁrst term leads to the Euler-Lagrange equations. ∂x ˙ (6.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar ﬁelds (Klein-Gordon equation). t1 (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. ˙ p= ∂L .

i.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. L (φ(x). x′ ) − d4 x L (φ. The resulting variation of the action is δS = R µ ′ = xµ + δxµ . ∂µ φ(x). . λ=0 (6. δ(∂µ φ) (6. t2 ). In general any continuous transformation. ˙ x The second term in Eq.2. . x). = ∂R (R3 . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x).18) 1 Taking a functional derivative. Variations in the action can come from the coordinates or the ﬁelds.13) (6.t2 ) d3 x .9) which is done because the ﬁrst term (multiplying ∆x. with ∆φ(x) = δφ(x) + (∂µ φ)δx . ∂µ φ′ . ∂µ φ(x)) = R d4 x L (φ(x). ∂R Note for the speciﬁc choice of the surface for constant times t1 and t2 .12) Here R indicates a space-time volume bounded by (R3 . but which in 3 dimensions includes conserved quantities related to rotations and boosts.e. indicated with δF [φ]/δφ should pose no problems. = φ(x) + δφ(x) (6. ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. x) ≡ p x − L .Classical lagrangian ﬁeld theory 45 For the lagrangian speciﬁed above this leads directly to Newton’s law p = m¨ = −∂V /∂x. x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). Consider a lagrangian density L which depends on these functions.10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). 6. ˙ (6. also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). their derivatives and possibly on the position.16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . (6. .14) (6. In classical ﬁeld theory one proceeds in complete analogy but using functions depending on space and time (classical ﬁelds.6 can be rewritten as δS = . . R (6.11) λ=0 of which p and H are the simplest examples related to space and time translations. think for instance of a temperature or density distribution or of an electromagnetic ﬁeld). . dσµ . ∂µ φ. ∂µ φ(x)). The hamiltonian H is deﬁned by H(p. + ∂L ∆x − H δτ ∂x ˙ t2 . . gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ . (R3 .t1 ) (6. which in classical mechanics vanishes at the boundary) does not play a role. . . t1 (6. H is a conserved quantity. t1 ) and R3 . We will come back to it in a bit more formal way in section 9. 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 .15) d4 x′ L (φ′ . . − d3 x . .

(2 + M )φ (x) = 0. δL δL ∂µ = . (6. 2 ∗ (6. (6. (6.19) δ(∂µ φ) δφ 6. ∂ (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).29) .20) (6. The integrand of the ﬁrst term must vanish.23) (6. For the complex scalar ﬁeld one conventionally uses L = = ∂µ φ∗ ∂ µ φ − M 2 φ∗ φ −φ∗ ∂µ ∂ µ + M 2 φ. Using the variations in ψ (in the symmetric form). One easily obtains (2 + M 2 )φ(x) = 0.25) (6.28) where the second line is not symmetric but in the action only diﬀers from the symmetric version by a surface term (partial integration).27) (6.22) which can be considered as the sum of the lagrangian densities for two real scalar ﬁelds φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. 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. leading to (2 + M 2 )φ(x) = 0. 2 ψ = 0. the Dirac ﬁeld (spin 1/2) and the vector ﬁeld (spin 1) can be found.2 Lagrangians for spin 0. the second term is irrelevant.24) (6. The scalar ﬁeld It is straightforward to derive the equations of motion for a real scalar ﬁeld φ from the lagrangian densities. leading to the Euler-Lagrange equations.Classical lagrangian ﬁeld theory 46 With for the situation of classical ﬁelds all variations of the ﬁelds and coordinates at the surface vanishing. 2 (6. L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 µ = − φ ∂µ ∂ + M 2 φ.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.

32) There exists another possibility to write down a mass term with only one kind of ﬁelds. The reasons for Grassmann variables will become clear in the next chapter.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ . (6. that the lagrangian separates into two independent parts for M = 0. L = = ↔ 1 1 c c M ψL ψL + M ∗ ψL ψL ψL i / ψL − ∂ 2 2 ↔ 1 1 M ΥR ΥL + M ∗ ΥL ΥR . (6.31) showing e.38) (6. The Majorana case is in fact more general.37) This gives an expression with a mass term that is actually of the same form as the Dirac mass term. and thus (β ∗ α)∗ = α∗ β = −βα∗ . In that case one ﬁnds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ).and lefthanded ones. ψL ≡ (ψL )c = C ψL = 0 (6. We note that of which the left part coincides . 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 . 2 (6.g. ∂ ∂ 2 2 (6. namely2 LM (Majorana) = + (6. Using the twospinors ξ and η.35) Since the kinetic term in L separates naturally in left and right parts (or ξ and η).g. ∗ 0 ⇒ Υ ≡ −ǫ η ΥL = ψL = η η ∗ for which η0 = ǫ η0 . or equivalently separate the ﬁeld into right.33) We now introduce a ’real’ (four-component) spinor satisfying Υc = Υ with ψL . 1 M η † ǫη ∗ − M ∗ η T ǫη .31 is given by LM (Dirac) = −M ξ † η + η † ξ . the mass term in the Dirac lagrangian 6. .36) (6.Classical lagrangian ﬁeld theory 47 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. but note the factor 1/2 as compared to the Dirac lagrangian. 2 αβ = −βα.30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4. Υi / Υ − ∂ 4 2 T −ǫ η ∗ c = ΥR . (6.34) (6. 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. 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.7). which comes because we in essence use ’real’ spinors. Peshkin and Schroeder or Exercise 12.

(6. .43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x). one can consider the variations at the initial or ﬁnal surface.44) δL ∆φ + Θµν (x)δxν .. + ∂R dσµ dσµ ∂R ≡ where d x .47) . In particular when the lagrangian is invariant under symmetries. 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. consider U (1) transformations of the Dirac ﬁeld proportional to the charge e.41) (6.45) J µ (x) = − δ(∂µ φ) dλ λ=0 dλ λ=0 Thus considering ∂R = σ2 . 6. Q(t1 ) = Q(t2 ). Q= d3 x J 0 (x. t). t). implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. . we get (omitting the parameter δλ) dxν dφ δL + Θµν (x) .σ1 with σ = (R3 . These do not aﬀect the dynamics and will give rise to conserved currents. In this case.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. which in a consistent picture precisely generate the symmetries. (6. (6. ∂µ J µ (x) = 0. discussed in the next chapter. δ(∂µ φ) (6. ∆φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ.3 Symmetries and conserved (Noether) currents Next. δ(∂µ φ) For continuous transformations. the presence of a symmetry that leaves the lagrangian invariant.Classical lagrangian ﬁeld theory 48 where Fµν = ∂µ Vν − ∂ν Vµ .42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . these conserved quantities become operators. we consider the second term in Eq. requires the presence of a conserved quantity.+ R 4 δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x). which is the space-integral over the zero-component of a conserved current. As an example for the classical ﬁeld.18.40) 6. (6.. (6. Returning to δS the surface term is rewritten to δS = R d4 x .

t) is the conserved charge (Q = 0). = 0. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )ij φj (x). including e the space-time translations and the Lorentz transformations. From Noether’s theorem one sees that the current Θµν ǫν is conserved. (6.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.g. For quantized ﬁelds this will become the expressions of the hamiltonian and the momentum operators in terms of the ﬁelds. (6. Summarizing.57) For the ﬁelds the behavior under Lorentz transformations has been the subject of previous chapters. 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.3) j µ = i e φ∗ ∂µ φ.56) δxµ = ωρσ (aρσ )µν xν 2 with (aρσ )µν = aρσ µν = g ρµ g σν − g σµ g ρν . (6.48) For the complex scalar ﬁeld the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6. 6. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . 2 (6. In the next section we will see the quantity show up as the charge operator. e.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations.54) (6. H = d3 x Θ00 (x). Lorentz transformations In this case the transformation of the coordinates and ﬁelds are written as δxµ ∆φi (x) = = ω µν xν . The integral over the zero-component. ↔ (6.49) These currents are conserved as discussed already in chapter 2. we obtain (since δxµ = 0). 1 (6. ˙ Q = d3 x j 0 (x.53) These are the energy and momentum.50) (6. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x).51) (6.55) Note that the coordinate transformations can be written in a form similar to that for the ﬁelds (matrices in internal space). (6. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ .

6. the angle of the transformation depends on the space-time point x. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current. e.64) (6.3). (6. • Dirac ﬁeld ψ: −i Sρσ = −(1/4)[γρ . however.65) M µρσ = T µρ xσ − T µσ xρ .58) (6. In general this tensor is not symmetric. is not invariant under local gauge transformations. in which the U (1) transformation involves an angle e Λ.5 (Abelian) gauge theories φ(x) → ei eΛ φ(x).3 we have seen global transformations or gauge transformations of the ﬁrst kind.63) µν = ∂µ Θ . Any lagrangian containing derivatives.67) i. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x). (6. 2 (6. γσ ] This result for the Dirac ﬁeld has been discussed in chapter 3 (see also Exercise 4. Deﬁning T µν = Θµν − ∂ρ Gρµν . (6. independent of x.59) (6.68) (6. −i eΛ(x) ∗ i eΛ(x) (6.66) In section 6. where H ρµν is the quantity appearing in M ρµν .e.Classical lagrangian ﬁeld theory 50 • Scalar ﬁeld φ: −i Sρσ = 0. φ(x) φ (x) ∂µ φ(x) ∗ → e → e → ei eΛ(x) φ(x).60) Therefore there are six conserved currents M µρσ labeled by ρ and σ (antisymmetric) and corresponding to it there exist conserved quantities M ρσ = d3 x M 0ρσ . • Vector ﬁeld Aµ : −i Sρσ = aρσ . one has a tensor that satisﬁes (exercise 6. φ(x). In some applications (speciﬁcally coupling to gravity) it is advantageous to have an equivalent current that is symmetric. (6. ∂µ T µν (6. 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 .g. ∂µ φ(x) + i e ∂µ Λ(x) e (6.62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2.61) A ﬁnal note concernes the symmetry properties of Θµν .6) T µν = T νµ .70) .69) i eΛ(x) φ (x). (6.

For this purpose it is necessary to introduce a vector ﬁeld Aµ . −e j µ Aµ = −e ρφ + e j · A. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x).72) (6. Dµ φ) − Fµν F µν .78) . (6. The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ.75) The ﬁeld φ is used here in a general sense standing for any possible ﬁeld.74) the behavior which we have encountered before as a gauge freedom for massless vector ﬁelds with the (free) lagrangian density L = −(1/4)Fµν F µν . (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ. (6. The required transformation for Dµ then demands Dµ φ(x) = = (∂µ + i eAµ (x))φ(x) ei eΛ ∂µ + i e(A′ + ∂µ Λ) φ µ → ei eΛ ∂µ φ + i e (∂µ Λ) ei eΛ φ + i eA′ ei eΛ φ µ ≡ ei eΛ (∂µ + i eAµ ) φ. 1 L (φ.76) We note ﬁrst of all that the coupling of the Dirac ﬁeld (electron) to the vector ﬁeld (photon) can be written in the familiar interaction form Lint = −e ψγ µ ψ Aµ = −e j µ Aµ . ∂ 2 4 (6.77) 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). D ∂ A (6. (6.Classical lagrangian ﬁeld theory 51 where the last term spoils gauge invariance. 4 (6. 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 . ∂µ φ) =⇒ L (φ. As an example consider the Dirac lagrangian. 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.73) (6. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisﬁes Dµ φ(x) → ei eΛ(x) Dµ φ(x). 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic ﬁeld.

give the equation which is satisﬁed by ψ c = Cψ . ↔ Exercise 6.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar ﬁeld is connected to a U (1) transformation on the ﬁelds.e.Classical lagrangian ﬁeld theory 52 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν .4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T ﬁeld. Exercise 6. giving the Maxwell equation coupling to the electromagnetic current. ∂µ F µν = j ν .2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν . 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. .1 (a) Show that the Klein-Gordon equation for the real scalar ﬁeld can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 . where j µ = e ψγ µ ψ.79) Exercises Exercise 6. i. −eψγ ν ψ. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved. and deduce what is the charge of the antiparticle as compared to a particle. ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). φ → eiΛ φ. 4 (b) What is the form of the interaction term involving a current jµ and the ﬁeld Aµ that will give the inhomogeneous Maxwell equations. Exercise 6. [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0. (6. ∂µ F µν = j ν .

5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic ﬁeld Aµ = (A0 .4. 2M 2M with Enr ≡ E − M ≪ M . Hint ≡ −gs µe · B = −gs Qe s · B.6 (optional) Prove the properties of the tensor T µν in section 6. in order to show that Θρσ − Θσρ = ∂µ H µρσ . Comment: this relation is known as the Pauli equation. ﬁnally note that it is suﬃcient to show the last equation for dσµ M µρσ .Classical lagrangian ﬁeld theory 53 Exercise 6. use ∂µ T µν = 0 and ∂µ M µρσ = 0.4) 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 . t) = Enr ψu (r. e 1 (−i∇ + eA)2 + σ · B − eA0 ψu (r. It shows that for a spin 1/2 particle the g-factor in the coupling of spin to a magnetic ﬁeld takes the value gs = 2. Hints: it may be useful to realize that Gµρσ is antisymmetric in ﬁrst two indices. . A) (see also exercise 6. This relation is known as the Pauli equation. 2me Exercise 6. t).

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

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

24) It is straightforward to ﬁnd the commutation relations between N and a and a† . [N. In simpliﬁed form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . a† ] = 1. ω a† a + 2 2 ω i − [Q.19) 7.17) (Sµν )i φj (x).20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. [Q. (7. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality).Quantization of ﬁelds 56 or if U (Λ. P ] = i. j (7. a† ] = 0.22) it is straightforward to check that the commutation relations between Q and P are equivalent with the commutation relations [a. (7. 2 2 (7. i(xµ ∂ν − xν ∂µ )φ (x) + i (7. The measurements cannot inﬂuence each other or [Φ(f ). Mµν ] = i i∂µ φi (x). Q] = [P.2 Creation and annihilation operators Before discussing (real and complex) scalar ﬁelds and Dirac ﬁelds we recall the analogy with the wellknown harmonic oscillator as an example of quantization using creation and annihilation operators. a] = [a† .23) The hamiltonian in this case can be expressed in the number operator N = a† a. sometimes referred to as second quantization. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. [a. j 2 one has [φi (x). Microscopic causality implies local commutativity. Φ(g)] = 0. P ] 2 (7. 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† ] = a† .21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators.4.18) with S µν given in 6. Pµ ] = [φ (x). 2 i ωµν (S µν )i . φ(y)] = 0 if (x − y)2 < 0. a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . P ] = 0 (7. (7. and [N.25) . a] = −a. [φ(x).

Note that we will often write a(k) or a† (k). (2π)3 2Ek (7. (n − 1) a|n † i. 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. a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k). H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . φ(x )]x0 =x′0 = [Π(x).e. † √ a and a act as raising and lowering operators.29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . [φ(x). but one needs to realize that in that case k 0 = Ek = k2 + M 2 . φ(x) = (2π)3 2Ek where the Fourier coeﬃcients a(k) and a† (k) are now operators.30) [φ(x).28) (7.26) a(k) e−i k·x + a† (k) ei k·x . which we have split into positive and negative energy pieces with (complex) coeﬃcients a(k) and a∗ (k) multiplying them. e.32) (7. and (7.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k).like’ commutation relations between a(k) and a† (k ′ ). N |n = n|n one ﬁnds N a† |n N a|n = = (n + 1) a† |n .27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. 7.33) . i. which represents the ’number of particles’ with momentum k. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.e. a† (k ′ )] = 0.4 for the classical ﬁeld) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). the real scalar ﬁeld φ(x) becomes d3 k (7.g. The quantization of the ﬁeld is achieved by quantizing the coeﬃcients in the Fourier expansion. is equivalent with [a(k). From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 . a(k ′ )] = [a† (k). Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.Quantization of ﬁelds 57 Deﬁning states |n as eigenstates of N with eigenvalue n. and we see that a state |0 must exist for which N |0 = a|0 = 0.3 The real scalar ﬁeld We have expanded the (classical) ﬁeld in plane wave solutions. (7. Π(x )]x0 =x′0 = 0. The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x .

which now contains an inﬁnite number of oscillators. (2π)3 2Ek (7.42) (7. .37) (7.e.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.41) (7. .35) where the vacuum contribution disappears because of rotational symmetry. which amounts to redeﬁning H as H= d3 x : H (x) := d3 k Ek N (k).40) This procedure is known as normal ordering. . 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). . is solved by subtracting it as an (inﬁnite) constant. deﬁning particle states |k = |k (with positive energy.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). d3 k |k k|.46) (7. |k = a† (k)|0 . For the purpose of normal ordering it is convenient to decompose the ﬁeld in positive and negative frequency parts. |0 . 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. (2π)3 2Ek d3 k a† (k) ei k·x . (2π)3 2Ek (7. (2π)3 (7.39) The problem with the vacuum or zero-point energy. k 0 = Ek = k2 + M 2 ).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). writing all annihilation operators to the right of the creation operators. i.38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7. n1 ! n2 ! (7. Just as in the case of the harmonic oscillator it is essential (axiom) that there exists a ground state |0 that is annihilated by a(k). = normalized as (a† (k1 ))n1 (a† (k2 ))n2 √ √ . and multiparticle states |(k1 )n1 (k2 )n2 .45) (7.44) p|φ− (x)|0 = ei p·x . This term will be adressed below. in ﬁeld theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek .34) where V = (2π)3 δ 3 (0) is the space-volume. a(k)|0 = 0. . d3 k a(k) e−i k·x . (2π)3 2Ek (7. The rest of the states are then obtained from the groundstate via the creation operator. assuring that the vacuum (by deﬁnition) has eigenvalue 0!. (2π)3 2Ek (7.

e.50) or as integrals over d4 k. The last item to be checked for the scalar ﬁeld are the causality condition. 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 (7. x) = 0 and hence ∆(x) = 0 for x2 < 0. = − (2π)3 2Ek (7. φ(y)] consider µν [φ+ (x).52) The result for the invariant commutator function is [φ(x).51) (7. (7. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.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.56) = −δ 3 (x).18).17) and (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . (2π)3 2Ek ≡ i∆− (x − y). i∆+ (x) i∆− (x) = = = d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . In order to calculate [φ(x). ∆(x) = − ǫ(x0 ) δ(x2 ). φ+ (y)] d3 k e−i k·(x−y) .54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . d3 k = (2π)3 2Ek = (7.53) t=0 . i. a† (k ′ )] 3 2E ′ (2π) k [φ− (x).47) ′ d3 k ′ e−i k·x+i k ·y [a(k). (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 (7.49) (7. (2π)4 (7. d3 k ei k·(x−y) = −i∆+ (y − x).48) (7. that they satisfy the required commutation relations in Eqs (7. (7. φ− (y)] ≡ i∆+ (x − y). (iii) ∆(0. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0.

63) i. (7. 2 From the lagrangian density (7. ↔ (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) . = ∂{µ φ∗ ∂ν} φ − L gµν .65) 7. we will consider it as a repetition of the quantization procedure. extending it with the charge operator and the introduction of particle and antiparticle operators. 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 . particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy.60) = i φ∗ ∂µ φ.66) . (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . φ† (y)] = i∆(x − y).59) (7.62) (7.57) (7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ). (2π)3 2Ek (7. which is equivalent to the relations (only nonzero ones) [a(k). (2π)3 2Ek (7.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) . (2π)3 2Ek (7.Quantization of ﬁelds 60 7.e.64) The commutator of φ and φ† is as for the real ﬁeld given by [φ(x). a† (k ′ )] = [b(k).61) (7.5 The Dirac ﬁeld L = ↔ i ψγ µ ∂µ ψ − M ψψ.4 The complex scalar ﬁeld In spite of the similarity with the case of the real ﬁeld.

(2π)3 2Ek (7.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . then leads to additional minus signs and H Q = = d3 k Ek b† (k)b(k) + d† (k)d(k) (2π)3 2Ek d3 k b† (k)b(k) − d† (k)d(k) . s)u(k.70) where the last line is obtained by using the Dirac equation. The solution is the introduction of anticommutation relations. . b† (k ′ . i.77) Note that achieving normal ordering.73) (7. u v (2π)3 2Ek (7. The quantized ﬁelds are written ψ(x) = s d3 k b(k.76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ.Quantization of ﬁelds 61 the conserved density and energy-momentum currents are easily obtained. s′ )} = {d(k. s)ei k·x . jµ Θµν = = ψγµ ψ. s). s)vj (k. ∂ 2 ↔ (7. s)e−i k·x .78) (7.79) Also for the ﬁeld and the canonical conjugate momentum anticommutation relations are considered.67) i ψ / ψ − M ψψ gµν . ↔ i ψγµ ∂ν ψ − 2 (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . {b(k. ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.71) (7. (7.e. (2π)3 2Ek (7.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x). s)v(k.74) (7. s)¯(k. s)ei k·x + d(k. (7. † {ψi (x). s) e−i k·(x−y) j s † + vi (k. interchanging creation and annihilation operators. s)u† (k. s) ei k·(x−y) x0 =y 0 . s)e−i k·x + d† (k. d† (k ′ . s)¯(k.69) (7. s). (2π)3 2Ek d3 k b† (k.75) (7.

It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . For the scalar combination {ψi (x). which however has wrong causality properties! Therefore the relation between spin and statistics is required to get micro-causality (which is also known as the spin statistics theorem).85) which reﬂects the gauge freedom.84) (7. When we would have started with commutation relations for the ﬁeld ψ and the canonical momentum.86) (7. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E i. 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 (7.6 The electromagnetic ﬁeld 1 L = − Fµν F µν . (7.83) From the lagrangian density the canonical momenta are = = (7. ˙ δ Ai (7. 7. 4 2 This gives the equations of motion discussed before.82) where i∆1 = i∆+ −i∆− . 4 Π0 Πi δL = 0.88) .Quantization of ﬁelds 62 Using the positive and negative energy projection operators discussed in section 4. we would have obtained [ψi (x). one has † {ψi (x). but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. ψ j (y)} at arbitrary times one has {ψi (x).81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before. −λ(∂ρ Aρ ). ψ j (y)] = (i/x + M )ij i∆1 (x − y). (7. ∂ (7. ψ j (y)} = d3 k (/ + M )ij e−i k·(x−y) + (/ − M )ij ei k·(x−y) k k (2π)3 2Ek d3 k e−i k·(x−y) + −ei k·(x−y) = (i/x + M )ij ∂ (2π)3 2Ek = (i/x + M )ij i∆(x − y).87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity.80) 3 = δ 3 (x − y) δij . as the vanishing of Π0 induces a constraint. but has the problem of being noncovariant. ∂ (7. ˙ δ A0 δL = F i0 = E i .

λ)c(k. 0) . µ λ=0 (7.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). pj ]P . where Q is the conserved quantity in Eq. 0)c(k.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. 2. 6. and are equivalent with [c(k. j = 1. one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies.11. (7. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. Q]P . 7.e. ′ (7. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. Aν (y)] = [Πµ (x). a longitudinal photon and two transverse photons. It gives 3 k µ ǫ(λ) (k)c(k. λ)ei k·x . λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ). (7. (7.92) 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.e. λ)e−i k·x + c† (k.94) Choosing k µ = (|k 3 |. [ℓi . These problems are solved by the Lorentz constraint between physical states given above.95) Exercises Exercise 7. ˙ ˙ . λ)|A = 0. µ λ=0 (7. pj ]P . and [ℓi . what are [ri . 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. i. q) that dO/dλ = [O. The quantized ﬁeld is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. λ) − c† (k.89) with four independent vectors ǫµ . Aν (y)]x0 =y0 = i gµν δ 3 (x − y). rj ]P . containing a time-like photon.91) (7. In fact the commutation relations imply ˙ [Aµ (x). The canonical equal time commutation relations are [Aµ (x). Πν (y)] = 0. where Aµ + + is the part of the vector ﬁeld containing the annihilation operators. (b) Proof for a quantity O(p.Quantization of ﬁelds 63 for physical states |A and |B . for which it is suﬃcient that ∂µ Aµ |A = 0. 0. which reproduce the Hamilton equations.93) which does exhibit problems with the time-like photon. c† (k ′ .3) satisfy the conditions of a Lie algebra. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). (d) Use Poisson brackets to calculate q and p. λ). i. 0.

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

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

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

˜ −v(k. λ) Pop = ηP b(k. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. λ) eik·x . −λ). λ) P −1 = −η ∗ d(k. x It is straightforward to apply this to the explicit ﬁeld operator ψ(x) using γ0 u(k. λ)v(k. For the scalar ﬁeld we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜). −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. . (i˜ µ ∂µ − M ) ψ(˜) = 0.8) 3 = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k.12) and for the vector ﬁeld −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜).10) i.7) = λ d k x x ηP b(k. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k. 8. one can also consider the scalar ﬁeld and vector ﬁelds.e. x (8.4) (8. λ) e−ik·˜ + d† (k. γ x (8.9) (8. the above operation reverses the sign of λ). (2π)3 2Ek (8.11) (8. λ)γ0 v(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0.6) (check this for the standard representation. op P (8. x The latter behavior of the vector ﬁeld will be discussed further below. −λ) e−ik·x − d† (k. In the same way as the Fermion ﬁeld. λ) e−ik·x − d† (k. −λ). x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). λ)γ0 u(k. −λ) e−ik·x − d† (k. λ)v(k. −λ)u(k. λ)u(k. iγ µ† ∂µ − M ψ(˜) = 0.13) ψ(x) −→ ψ c (x) = ηC C ψ (x). m). m). m) γ0 v(k. λ)u(k. −λ)v(k. ˜ Pop d(k.3) (8. choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity. m) = = ˜ u(k.5) (8.Discrete symmetries 67 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0.2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. if helicity λ is used instead of the z-component of the spin m.

λ) −1 Cop = ∗ ηC b(k. u(k. m) ¯ = = v(k.18) (8. x (8. For a quantized ﬁeld one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. antilinear operator is anti-unitary if A† = A−1 . as discussed in section 4). Norm conservation requires Top to be anti-unitary2 . m) ¯ C v T (k. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. r) −→ −˜µ ≡ −xµ = (−t. x iγ µT ∂µ − M ψ(−˜) = 0.20) The time reversal operator transforms We will again consider the transformation properties for a fermion ﬁeld ψ(x).e. λ) e−ik·x + b† (k.19) Cop d(k. λ)C v T (k.21) where A is a 4 × 4 matrix acting in the spinor space. λ)u(k. The same relations hold for helicity states. r). λ)v(k. Top |φ = φt | = (|φt )∗ . λ)C uT (k. As time reversal will transform ’bra’ into ’ket’. λ). to ﬁnd ψ t (−˜) that is a solution of the Dirac equation.3 Time reversal xµ = (t. x (8. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0.16) d k ηC d(k.17) 3 = λ This shows that −1 Cop b(k. i.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = λ∗ A|φ + µ∗ A|ψ . λ) e−ik·x + b† (k. we start with the complex conjugated x Dirac equation for ψ. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.14) (8. m). writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). (8. (8. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. 8. (2π)3 2Ek (8. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. One has Aφ|Aψ = φ|ψ ∗ = ψ|A† Aφ = ψ|φ . λ) eik·x . it is antilinear1 . (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. The (time-reversed) Dirac equation becomes. x x (8. m). . ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. λ) Cop = ηC d(k. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. λ).15) (where one must be aware of the choice of spinors made in the expansion.Discrete symmetries 68 the latter being also a solution of the Dirac equation.

29) In table 2 the behavior of particle states under the various transformations has been summarized. λ). λ). λ) T −1 = η ∗ d(k. λ) e−ik·x 3 2E (2π) k ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k. λ) i γ5 C v(k. λ) e−ik·x (2π)3 2Ek (8. there are 16 independent of these bilinear combinations.34) . p. p. λ) eik·˜ + d† (k. λ). λ)v ∗ (k. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. ∗ ˜ v (k.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor ﬁelds are encountered. λ) eik·x + d† (k. λ)u∗ (k. −p. −λ |¯. Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. a state |a. (8.25) (check this for the standard representation). λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). λ)v ∗ (k. λ a P |a. p.33) (8. x (8. Since there are 16 independent 4 × 4 matrices. −p.30) (8. The result is = = λ (8. op T (8. λ)u∗ (k.32) (8. λ) e−ik·x 3 2E (2π) k d3 k ˜ ˜ ηT b(k. ˜ Top d(k. λ). −λ a T a. λ a |a. ˜ = 0|b(k)|A 8. −p. λ)v ∗ (k. λ) eik·x + d† (k. λ)u∗ (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. λ| a.24) (8. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) ¯ = ψ(x)γ µ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) µ ¯ = ψ(x)γ5 γ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ)iγ5 Cu(k.Discrete symmetries 69 Table 8.2: The transformation properties of physical states for particles (a) and antiparticles (¯). −p. p.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit ﬁeld operator ψ(x) using i γ5 C u(k. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ) Top = ηT b(k.31) (8. λ) eik·x + d† (k.28) (8. λ| ¯ C |¯. λ |¯. λ)iγ5 Cv(k.26) d k x x ηT b(k.

p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ).37) (8. p)u(p) e−i (p−p )·x . ¯ In general the expectation value between momentum states can be more complicated. and T. σ µν . V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . γ5 γ µ or γ5 ). p) can be built from any combination of Dirac matrices (1. 8. as well as under the combined operation Θ = PCT (see Table 3). p′ . γ µ . As an example consider the vector current for a point fermion. F. F. called form factors.g.3: The behavior of the independent bilinear combinations of fermi ﬁelds under P. Note that the lagrangian density L (x) → L (−x) under Θ. The 16 combinations of Dirac matrices appearing above are linearly independent.39) where Γµ (p′ . ¯ ′ ′ (8.35) The x-dependence can be accounted for straightforwardly using translation invariance.38) (8. This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . 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 . in this case only q 2 = (p′ − p)2 . of which the expectation value between momentum eigenstates can be simply found.Discrete symmetries 70 Table 8. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P. p′ |V µ (0)|p = u(p′ )γ µ u(p). (q 2 ). C.. 2M (8. There are several other terms that also have the correct tensorial behavior such as (pµ + p′µ ) γ5 σ µν qν q µ F3 (q 2 ). momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). C. For instance for nucleons one has Γµ (p′ . p′ |V µ (x)|p = u(p′ )Γµ (p′ . for the vector current V µ (x). In many applications the expectation values of currents are needed. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :. e.5 Form factors Currents play an important role in ﬁeld theory. s′ |V µ (x)|p.41) 2M 2M . s . γ ν ]. (8. (q 2 ).. (8.40) where the coeﬃcients are functions of invariants constructed out of the momenta.36) (8. As the coupling of the photon ﬁeld to fermions is given by an interaction term in the ¯ lagrangian of the form : ψ(x)γ µ ψ(x) : Aµ (x) and behaves as a scalar one sees immediately that the ¯ photon ﬁeld Aµ (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x).

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

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

(9. t0 ). t0 )|ψS (t0 ) . t0 ) AH (t0 ) U (t.6) Of these the Heisenberg picture is most appropriate for quantum ﬁeld theory since the ﬁeld operators do depend on the position and one would like to have position and time on the same footing.4) ∂ U (t. t ′ QH (t′ )|q ′ . t0 ) = H U (t. t ≡ q|q. and the operators are time-dependent. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS .Chapter 9 Path integrals and quantum mechanics 9. ≡ 0. t′ ≡ q ′ |q ′ .5) (9.1 Time evolution as path integral U (t. t . t ′ |q . |ψH (t) = |ψH . i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. |ψS (t) = U (t.2) (ii) Heisenberg picture. [AH . AH (t) = U −1 (t. H].1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian. (9. in which the operators are time-independent.3) (9. t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture. in which the states are time-independent. t′ . t0 ) ∂t (9.7) 73 . (9. QH (t)|q. AS (t) = AS and the states o are time dependent. Consider the two (time-independent) Heisenberg states: |q. (9.

t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ . Then qj+1 .15) By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj . tj+1 |qj . t′ |q. n |q (9.9) 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 ′ . with the correction1 being of order (∆τ )2 . Q) expressed in terms of the operators P and Q. . QS |q ′ ≡ q ′ |q ′ . The hamiltonian is an operator H = H(P. These can be written in coordinate or momentum representation as Q = P = dq |q q q| dq |q −i ∂ ∂q q| = dp |p p p|. B] + .16) this. The purpose of this is to calculate the evolution for an inﬁnitesimal time interval. qj )]) . (9.. [A. (9. qj ). Combining the two terms gives qj+1 .qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj . H(P. B].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 . . |q1 q1 |e−i H∆τ |q .11) (9.Path integrals and quantum mechanics 74 and Schr¨dinger states o |q |q ′ QS |q ≡ q|q . tj+1 |qj .q .14) . eA eB = eC with C = A+B+ (9. ˙ 2π 1 1 1 [A. Q) = K(P ) + V (Q) = (P 2 /2M ) + V (Q) one can split e−i H∆τ ≈ e−i K∆τ e−i V ∆τ .qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj . tj = = = 1 For dpj i pj (qj+1 −qj ) −i H(pj . . B]] + [[A. 2 12 12 (9.10) dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . 2π (9. use the Campbell-Baker-Hausdorﬀ formula.8) Choose t as the starting point with |q = |q.12) where the transformation between coordinate and momentum space involves q|p = ei p.. t = q ′ |e−i H(t −t) |q . t′ |q. 2π (9. ′ (9. which is a hamiltonian function in which the operators are replaced by real-numbered variables. t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 . B] + [A.

dx dy α∗ (x) K(x. i. α(x) → αx and F [α] = F (αx ) changes into a multivariable function.e. 9. The importance of this expression is that it expresses a quantum mechanical amplitude as a path integral with in the integrand a classical hamiltonian function. . qN ) ˙ . dx dy α(x) K(x. . q)] ˙ (9. . . including diﬀerential operators. q1 .17) = Dq D dτ [pq − H(p. . F [α] ∈ R. dx α2 (x). while Dα F [α] = x dαx N F (αx ) (9.20) becomes a multidimensional integral. ≡ exp − 1 2 dx α2 (x) . q)] . .e. . . i. Before proceeding we also give the straightforward extension to more than one degree of freedom. What one is after is the meaning of Dα F [α]. β] = α K β ≡ 1 F4 [α] = exp − α2 2 F5 [α. y) in the above examples can be (hermitean) operators acting on the functions. qN . y) α(y) .Path integrals and quantum mechanics 75 or for the full interval q ′ . . etc. Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells. . Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x). . α∗ ] = exp (−α∗ K α) ≡ exp − The kernels K(x. F3 [α. qN .18) t ′ = n=1 Dqn D pn 2π exp i t dτ n pn qn − H(p1 . ˙ where Dq and Dp indicate functional integrals. . 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.2 Functional integrals In this section I want to give a fairly heuristic discussion of functional integrals. t N (9. (9. y) β(y). t′ |q. . t = τ dq(τ ) dp(τ ) exp i 2π p exp i 2π t′ t t′ t ∆τ [pq − H(p. t′ |q1 . ′ ′ q1 .

25) (see Exercises).27) (9. A useful property of functional integration is the translation invariance. Using method 1 one writes 1 Dα exp − α2 2 = x dαx N exp − 1 2 ∆x α2 x x = x dαx 1 exp − ∆x α2 x N 2 1 N 2π ∆x ≡ 1.n fm (x) Kmn fn (y) and consider F [α] = F (αn ) as a multivariable function. with Dα F [α] = n dαn N F (αn ) (9. α(x) = n αn fn and K(x. (9.22) = x The last equality is obtained by deﬁning the right measure (normalization N ) in the integration.26) As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω .Path integrals and quantum mechanics 76 (ii) Write α in terms of a sum of orthonormal basis functions. 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. y) = ∗ m.23) Having deﬁned the Gaussian integral.21) again a multidimensional integral. Note that procedure (i) is an example of the more general procedure under (ii).28) Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) . Dα F [α] = Dα F [α + β].n dx fm (x)fn (x) = n = n dαn 1 exp − α2 N 2 n √ 2π ≡ 1. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. Consider the gaussian functional as an example. 2 (9. the following integrals can be derived for a real symmetric or complex hermitean kernel K. .24) (9. In both cases N is an appropriately choosen normalization constant in such a way that the integral is ﬁnite. N (9. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 = n dαn N exp − 1 αm αn 2 m. (9.

θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is deﬁned as dθ 1 = 0 and dθ θ = 1.38) . however. y) β(y) δα(x) δ 1 1 exp − α2 = −α(x) exp − α2 . dx dy θ∗ (x) K(x. the Gaussian-type functionals. (9. Examples of functional diﬀerentiation are (9. θη = −ηθ. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. δα(x) (9. (9. which vanish. θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ. δα(x) δα(x) δ αKβ = dy K(x. δα(x) or in discretized form δ 1 ∂ ∂ with . products of the same pair appear.e. δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) .29) (9. θ2 = 0. αy = δxy . F [θ.32) (9.Path integrals and quantum mechanics 77 Functional diﬀerentiation is deﬁned as δ .g.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. (9. In the expansion of the exponential (from second to third line). n θn fn (x) dx θ∗ (x) θ(x) . αn = δmn . y) θ(y) . e.30) (9. i.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. we need to consider Grassmann-valued functions involving anticommuting Grassmann variables. α(y) = δ(x − y). −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) = n ∗ fn (x) fn (y). In principle the deﬁnitions of functionals is the same. −→ δα(x) ∆x ∂αx ∂αx δ ∂ ∂ ∗ with .34) (9.33) (9.31) δ δ α = dy α(y) = dy δ(x − y) = 1.35) For applications to fermion ﬁelds.

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

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

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. 9.6 Euclidean formulation The above expression (Eq. δJ(tn ) hence the name generating functional.67) = = φ0 (q. t) (9. Since we have (neglecting multiplicative factors).Path integrals and quantum mechanics t 82 T t t’ T’ Figure 9. t′ ) q ′ .70) for the generating functional is in fact ill-deﬁned. T J lim −i E0 (T ′ −T ) φ (Q′ ) φ∗ (Q) 0 0 T ′ → −i∞ e T → i∞ dq ′ dq φ∗ (q ′ . = (i)n 0|T Q(t1 ) . dq dτ + J(τ )q(τ ) .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 ′ ). J=0 (9. J (9.68) (9. (9. in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. Q(tn )|0 . such that t′ → −i∞ t → i∞ ↔ ↔ ˜ t → −∞.71) 9. t). . t 0 0out |0in J . eiE0 T φ∗ (Q) 0 We deﬁne the generating functional Z[J] as Z[J] = Q′ .70) one immediately ﬁnds δ n Z[J] δJ(t1 ) . Better is the use ˜ of imaginary time t = −it. T ′ |Q. T → i ∞ and T ′ → −i ∞. t′ |q. thus T →i∞ lim q. . . . T = φ0 (q.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time. The importance of Z[J] is that the time ordered product of operators can be obtained from it. ˜ t′ → ∞. T′ Z[J] = lim T ′ → −i∞ T → i∞ Dq exp i T dτ L q.

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

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

t′ |φ. δJ(xn ) 85 ≡ G(n) (x1 . .3) The Euclidean formulation is as before.Chapter 10 Feynman diagrams for scattering amplitudes 10. implying at the level of four vectors in coordinate and momentum space (when k · x ≡ kE · xE ) for instance x4 = i x0 = i t x · x = (x0 )2 − d4 xE = i d4 x L = 1 2 3 i 2 i=1 (x ) k 4 = −i k 0 =− 4 µ 2 µ=1 (x ) k · k = (k 0 )2 − d4 kE = −i d4 k 3 i 2 i=1 (k ) =− 4 µ 2 µ=1 (k ) ∂µ φ∂ µ φ − 1 2 M 2 φ2 LE = 1 2 µ (∂E φ)2 + 1 2 M 2 φ2 Furthermore we ﬁnd the n-point Green functions 0|T φ(x1 ) .1) 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. φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) . i.4) J=0 . ∂µ φ) + Jφ] (10. xn ).e. . . ∂µ φ) − Jφ] (10.1 Generating functionals for free scalar ﬁelds The generating functional for quantum ﬁelds is a generalization of the results in the previous section to a system with more degrees of freedom. t = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ. x. . φ(x. (10. . . . t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ′ . Π) (10.2) = ∞ −∞ d4 xE [LE (φ. x′ . .

d4 xn G(n) (x1 . = 0 2 2 2 k2 − M 2 (k ) − E 2 (k ) − k − M (10. As discussed in the previous section (exercise 9.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. . . 2 2 → (10. Furthermore. consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x). one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10. (10.13) .9) i.10) J=0 In order to determine ∆F . (2π)4 (k 0 + iǫ)2 − E 2 (10. (10. .12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10. with Z0 [0] = 1. . .5) Z0 [J] for the free scalar ﬁeld For the (free) scalar ﬁeld lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . xn ) J(x1 ) . . xn ) = x2 xn . (10. which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = 1 1 1 = 0 2 .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). .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. . . .e.7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). These Green functions appear in the expansion Z[J] = n in n! d4 x1 . . J(xn ).3) one ﬁnds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) .Feynman diagrams for scattering amplitudes 86 for which we introduce the picture x1 G(n) (x1 . .8) 1 d4 x φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K (10.

φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x). The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space. ∆(x) = − d4 k e−i k·x . t) = 0 for t < 0. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). i.Feynman diagrams for scattering amplitudes 87 satisfying ∆R (x. (2π)4 k 2 − M 2 (10. (10.16) C where C is a closed contour in the complex k 0 -plane.18) Thus we see various solutions. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (10.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. (2π)4 (k 0 − iǫ)2 − E 2 (10.15) e−i k·x d4 k . (10. 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 . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). 10.22) (10. ∆F = ∆C .17) d4 k e−i k·x . (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . 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). 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).19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) . φ+ (0)] = (2π)3 2E are also shown in exercise (7. 4 k 2 − M 2 + iǫ (2π) (10. The ﬁrst possibility is to consider the well-deﬁned Euclidean formulation.14) They are solutions of the inhomogeneous equation.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10.20) = exp − 2 where or (see ﬁg. . t) = 0 for t > 0. µ 1 1 • Firstly.e. (2π)4 k 2 − M 2 (10. In order to see which is the appropriate Green’s function to be used in the generating functional Z0 [J] we will take two routes. the advanced Green’s function ∆A (x) = satisfying ∆A (x.23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 .4).

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

x2 ) = i ∆F (x1 − x2 ) = G(4. x2 . L (φ) = L0 (φ) + LI (φ). with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].32) (10.g 0 ) = G0 = 1.34) Introducing a vertex point to which four propagators are connected.g ) (x1 .g ) (x1 . 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 .30) and in zeroth order G(0.35) z . To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . x3 . Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar ﬁeld theory discussed sofar.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) (10.28) 1 exp − i 2 = exp 1 2 d4 x d4 y exp i δ δ i ∆F (x − y) δφ(x) δφ(y) d4 z [LI (φ) + J(z)φ(z)] φ=0 . ≡ −i g d4 z (10.31) G(2. i.27) When interactions are present.Feynman diagrams for scattering amplitudes 89 10.29) This expression will be the one from which Feynman rules will be derived. x3 . x4 ) = To ﬁrst order in g one has Z (g ) [J] = = 1 0 0 x1 + + x2 (10. (10. (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. x4 ) = G0 (x1 . (10. (2) (0) (10. the generating functional can be written.e. Z[J] = = Dφ exp i exp i d4 z LI (10. x2 ) = G0 (x1 .

This remains also true for the interacting case. (10.g ) (x1 . . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . i. J(xn ) c (10. did not contain parts that could be written as products of simpler Green’s functions.42) x1 G(n) (x1 . .40) in−1 n! d4 x1 . xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . In the free case we have seen that the exponent of Z[J] contained all essential information. 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]. . . . .41) in terms of socalled connected Green’s functions. G(n) (x1 . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . Diagrammatically this exponent only contained the two-point function. J(xn ). . xn ) = x2 c xn Speciﬁc examples are G(0) = c G(1) (x1 ) = c G(2) (x1 . δJ(xn ) . .g 1 ) = 1 8 z 1 8 .36) J J from which one obtains G(0. (10.Feynman diagrams for scattering amplitudes 90 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. . d4 xn G(n) (x1 .37) G(2. (10. . (10.e. . . . . . J=0 (10. xn ) J(x1 ) . .38) The result for the 4-points Green’s function is left as an exercise.39) We have calculated the function up to ﬁrst order in the coupling constant for φ4 -theroy. . . . 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]. . which also was the only Green’s function for which the diagram was connected. d4 xn G(n) (x1 . (10. . . . . . xn ) J(x1 ) . Z[J] = n in n! d4 x1 .

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

g ) (x1 . implying the absence of tadpoles. Sβα ≡ β. out| ⇔ |α. in|SS † |α. out|p′ . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). this will be translated to the action on ﬁelds. pn . thus β. x4 10.47) . out| = eiϕ0 0.g. in = p. Proof: α. p′ . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider ﬁelds φin . e. out|α. (10. out = p|p′ . in into ﬁnal state particle states (out-states) |β. in| (choose ϕ0 = 0). then (n) (n) Gsc = Gc for n ≤ 3.45) x2 as the only surviving diagram (see exercises). (10. p. For the free scalar theory.e.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 ). φout and the interpolating ﬁeld φ(x). in ⇔ β. one ﬁnds for the connected 4-point Green’s function at order g.44) J=0 For the interacting theory. . translation invariance). in|S|p′ . out . out . 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 . Proof: 0. out = |p′ . (3) S is unitary (it conserves the scalar product from initial to ﬁnal state). in|p′ . . (2) The one-particle state is invariant (conservation of energy and momentum. out|p′ .Feynman diagrams for scattering amplitudes 92 i. considered explicitly. x1 G(4. in|S = β. ∂µ ∂ µ + M 2 φin (x) = 0. out (suppressing except momenta all other 1 m quantum numbers). x2 . in = S|α. in = β. (2) (10. out|α. .3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. in|S = 0. in = p. out| and S † |α. the vacuum expectation value of the ﬁeld φ(x) is zero. in|S = α. . . x4 ) = c 1 x3 z (10. x3 .46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. Next. . (10. in = |α. in = p. Also for free ﬁeld (Heisenberg) operators a distinction is made between φin and φout . out = δαβ ↔ SS † = 1. in = |p1 .

z) φ(z).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). φ(y)] = Z [φin (x). (2π)3 2E (10. Proof: β.49) The ﬁeld φ can be expressed in φin/out using retarded or advanced Green’s functions.54) . φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (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 ). in|φin S → φin S = S φout (x). √ Although the above. the source term Jφ in the lagrangian density is treated in the interaction picture.62) is then T′ UT [J] T′ = T exp i T d4 x J(x)φ(x) . The relation between S and Z[J] To establish this relation. which in a later stage (renormalization) will become more important.and out-ﬁelds is: φin (x) = S φout (x)S −1 . z) φ(z) d4 y d4 z ∆A (x − y) K(y. e Proof: φin (Λx + a) = U φin (x)U −1 = U Sφout (x)S −1 U −1 = U S U −1 U φout (x) U −1 U S −1 U −1 = U SU −1 φout (Λx + a) U S −1 U −1 ⇔ U SU −1 = S. e. i. The relation between S-matrix and in. 2 ∂µ ∂ µ + M0 φ(x) = J(x). (10. The time evolution operator (see Eq. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10. (10. as it stands. a causality condition that can be proven to imply the absence of interactions.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y. a)SU −1 (Λ. this can actually not be used as it would imply [φ(x). 9. S is invariant under e Poincar´ transformations: U (Λ. 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.51) (10. The convergence therefore must be weakened to t→−∞ √ (10.48) 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. (10.e.52) where Z is a constant. φin (y)] = iZ ∆(x − y). implies the strong (operator) convergence φ(x) → Zφin (x).g. β. a) = S. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance. out|φout = β.

(10. e = i. S U [J]] = √ Z δ S U [J]. express it in terms of advanced and retarded Green’s functions.56) Since δU [J]/δJ(x) satisﬁes the same equations as φ(x) and we know the limits. y) δ δJ(y) : F [J]. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. (10. φ(x).57) i [φin (x). B] is a c-number). : e d4 z φ(z) f (z) d4 x φ− (x) f (x) B C B C (10.61) e φ+ f where the normal ordered expression : e φ f : is used. z) d4 y d4 z ∆(x − y)K(y. just as we did for φ(x).59) (10. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. we can. B]eB . Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J]. t i Ut∞ [J] φ(x)U−∞ [J] (10. B] and [A. z) .62) where F [J] is some (arbitrary) functional. δJ(z) (10. [A. C] are c-numbers. δJ(z) (10. one has [A.Feynman diagrams for scattering amplitudes 94 and satisﬁes the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. 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(x). provided [A.55) δU [J] δJ(x) = t→∞ t→−∞ −→ −→ √ i φ(x) U [J] → i Z φout (x) U [J] √ i U [J] φ(x) → i Z U [J] φin (x). φ(y)] f (y) e e . 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. Thus applied to the ﬁeld φ(x) = φ+ (x) + φ− (x).e.63) . with the use of the Baker-Campbell-Hausdorﬀ formula e−B A e+B = A + [A. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x.60) e d4 y φ+ (y) f (y) d4 z φ− (z) f (z) d4 z φ+ (z) f (z) d4 y [φ(x). B] (for the case that [A.58) In order to ﬁnd a solution to this equation note that. z) δU [J] . e e ] = [A. which is equal to the expression e in which the creation operators are placed left of annihilation operators. eB ] = [A. : =i d4 y ∆(x − y) f (y) : e d4 z φ(z) f (z) :. φ− f (10. B + C]e e .

. . . iKx′ . (λ) v(p). J=0 (10. ←− − → × G(n) (x′ . pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . . 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). . which is obtained considering only connected diagrams. . y). fp′ (x′ ) . Usually we are interested in the part of the S-matrix describing the scattering. .Feynman diagrams for scattering amplitudes 95 while for J = 0 one has U [0] = 1. |S|p. (2π)3 2E n (10. They start with the propagator (i∆F (k)) in momentum space. . . . n (2π)4 δ 4 (p1 + .64) Therefore. .66) where Gc is the connected Green’s function. .2). i. i. .) iKx . . p′ . . . . c (n) (10.e. . . y) δ δJ(y) : Z[J] Z[0] . . .. which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. . see exercise 10.68) 10. k = k2 i − M 2 + iǫ .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. + pn ) G(n) (p1 .65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula. .67) where ∆(p) is the Fourier transform of the (full) propagator ∆F (x). . . It is straightforward to check that the S-matrix element now precisely is given by the amputated Green’s function multiplied with the momentum space wave functions of the particles in initial and ﬁnal state (by which we refer to the quantities multiplying the plane wave e±i p·x in the ﬁeld expansion. .e. u(p). S =: exp 1 √ Z d4 x d4 y φin (x) K(x. .. . (10. fp (x) . . . . . ∆(pj ) j=1 (10. pn ).. = . ¯ ¯ and the amputated or one-particle irreducible (1PI) Green’s functions n −i (n) Γ(n) (p1 .+ 1 √ Z ∗ d4 x′ . . d4 x . using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . . xn ). u(p) and v (p) for fermions and ǫµ (p) for vector ﬁelds). . Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function.e. pn ) = G (p1 . . . . Explicitly. Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance. . 1 for scalar case. .. .δ. . . . x.

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.27. 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. ’t Hooft and M. or calculating full Green’s functions external lines are treated as propagators. For the interaction terms in the lagrangian. vertices and external particle wave functions in that diagram.66 and 10. 3! 4! The vertices are: = −i f Consider ﬁrst the lowest order self-energy diagram. If problems arise go back to the deﬁning expression for the generating function in 10. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically diﬀerent connected diagrams. For the symmetry factor consider the examples (given in G. Diagrammar) in the case of the interaction terms g f (10.Feynman diagrams for scattering amplitudes 96 (in fact the inverse of the operator found in the quadratic term is for real scalar ﬁelds also multiplied by a factor 2. the factor 2 corresponds to the two-points Green’s function having two identical ends).27. but overall momentum conservation at a vertex is understood. Note that in calculating amputated Green’s functions external lines are neglected. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). = −i g . to be precise iLI vertices in momentum space are introduced. At these vertices each line can be assigned a momentum.69) LI (φ) = − φ3 − φ4 . Veltman. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines ﬁxed). which cancels the factor 1/2 in the quadratic piece.

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

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

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

s2 . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ .s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 . s′ ) ¯ u s. t2 (10. s1 ) ¯ −igµν u(p′ . s3 )γ µ u(k.s4 e4 (m) µν (M) L L . s2 )|2 ¯ q2 (10. s4 )(−ie)γ ν u(p. s4 )γµ u(p.5 Some examples eµ scattering The ﬁrst example is the electromagnetic scattering of an electron and a muon. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. s)γν u(k ′ . ¯ q2 (10. s)¯(k. s′ )γµ u(k. where u ¯ L(m) µν = = 1 2 u(k ′ .75) Note that the q µ q ν term in the photon propagator are irrelevant because the photon couples to a conserved current.Feynman diagrams for scattering amplitudes 100 10.77) 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. s2 ).s3 .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s |¯(k ′ . s1 ) u e2 u(p′ .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 ) (10.79) . 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 .76) 1 .78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 . To lowest order (∝ α = e2 /4π) only one diagram contributes. s3 )(−ie)γ µ u(k.

Exercises Exercise 10. s4 ) u e2 v (k ′ . s4 ) v (p′ . s2 )γµ u(k. s4 )γ ν u(p. (c) Do the same for the connected Green function Gc . s3 )γ µ v(p′ .s4 = 1 1 e Tr(/ − m)γµ (/ ′ + m)γν k k Tr(/ + M )γ µ (/′ − M )γ ν p p s2 2 2 1 e4 1 ′ ′ = 4 2 kµ kν + kν kµ − gµν s pµ p′ν + pν p′µ − g µν s s 2 2 M 4π 2 4 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . s2 )γµ u(k. (4) (4) .s2 3 . (b) Use the deﬁnition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive. s3 ) ¯ ¯ 1 .80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . The diagram. s2 (10. the masses. momenta and invariants are e− e+ → µ− µ+ scattering k p s t = = (k + k ′ )2 = (p + p′ )2 (k − p)2 = (k ′ − p′ )2 (k − p′ )2 = (k ′ − p)2 k’ p’ u = The scattering amplitude squared (spins summed and averaged) is given by |M |2 = 1 4s e4 s2 |¯(p. .s3 .s4 = 1 2s × v (k ′ .1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . s3 )γ µ v(p′ . Similarly.Feynman diagrams for scattering amplitudes 101 The second example is the annihilation of an electron pair and creation of a muon pair. . s1 ) u(k. x4 ) for the interacting case in φ4 -theory to ﬁrst order in the coupling constant g as obtained from the full expression for Z[J] in section 10. for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. This is known as crossing symmetry.2. s2 ) ¯ ¯ 1 2s u(p. . To lowest order (∝ α = e2 /4π) only one diagram contributes.s2 . Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. s1 )γν v(k ′ . s1 )|2 ¯ s 1 .

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

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

p2 ) = = where λ12 denotes λ(s.8) (11. 4s (11. m2 . m2 ) is a function symmetric in its three arguments. m2 . m2 .10) The function λ(s. ptrf ). 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. 2 s s − m2 + m2 a b √ . which in the speciﬁc case a b also can be expressed as λ(s. In that case trf pa = (Ea . m2 ) a b . m2 .Scattering theory 104 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma .15) Explicit calculation of the reduced two-body phase space element gives dR(s. p1 .12) pb = (mb . Prove that trf Ea = |ptrf | = a s − m2 − m2 a b . 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.9) (11.11) (11. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. m2 ) = 4(pa · pb )2 − 4p2 p2 . 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 √ = . a (11. = 2 s λ(s. m2 . p1 . 2mb λ(s.14) One can. Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = .13) (11. 2mb (11. for instance. m2 ) a b . mb and s). 4π s 4π 8π s 4π (11. (11.16) . Also in this case one can express the energy and momentum in the invariants. 0). trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . m2 ).

however. An often used set of invariants are the Mandelstam variables. .4). the rapidity y is deﬁned T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz .24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. E . a c with q = λab /4s and q ′ = being 0 or 180 degrees.20) → π0 + n → π+ + π− + n → .. The initial state.25) dR(s. pc . dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. there is usually a preferred direction. 11.22) (11.21) (11. = (11. The variable s is always larger than a c b d the minimal value (ma + mb )2 . py ). A speciﬁc reaction channel starts contributing at the threshold value (Eq. pz = mT sinh y. At high energies. 2s (11. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . The various possibilities are referred to as diﬀerent reaction channels. 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 ..18) (11. while the ﬁnal state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). (11. In inclusive measurements no particles are detected in the ﬁnal state. Consider the process a + b → c + d.26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable.17) (11. Consider the 1-particle inclusive case. 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 . in which one particle is detected. pd ) = 4π s 4π 8π s 4π dt dt √ √ . usually is a 2-particle state. in exclusive measurements all particles are detected. H1 + H2 → h + X.27) with m2 = m2 − p2 = m2 + p2 + p2 . Writing E = mT cosh y. pT = (px .19) (11. usually many particles are produced.Scattering theory 105 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out. where the ﬁrst is referred to as elastic channel and the set of all other channels as the inelastic channels. Examples appear in → π− + p (11. tmax min λcd /4s. Instead of the scattering angle.

t. ¯ b (11. b Analiticity of the ﬁeld theoretically calculated result implies Ma¯→¯ (st = t. tu = s.2 Crossing symmetry In the previous chapter.28) which (only involving angles) is easier to determine. we have seen that the amplitudes for the processes e− µ− → e− µ− and the process e− e+ → µ+ µ− are simply related by an interchange of the variables s and t. t.Scattering theory 106 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. t-channel and u-channel processes respectively. u). uu = (pa − p¯)2 = s. u). η = − ln(tan(θ/2)) = tanh−1 (cos θ). This is known as crossing symmetry.29) 3 2E 3 2 (2π) 2 (2π) 16π 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). The one-particle phase space in terms of these variables becomes d3 p dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ = = = . ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u.31) . For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. With the momenta deﬁned as in the ﬁgures above one has for all these processes the same amplitude M (s. ut = u) = Mab→cd (s. t. t = (p1 − p3 )2 = (p2 − p4 )2 . Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. u = (p1 − p4 )2 = (p2 − p3 )2 .30) (11. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . (11. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η. 11. ¯ tu = (pa − pc )2 = t. b ut = (pa − pd )2 = u. (11. tt = s. which means that rapidity distributions dN/dy maintain their shape. uu = s) = Mab→cd (s. c bd Mad→c¯(su = u. ¯ tt = (pa − p¯)2 = s.

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

N and T · V are covariant. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. . T · V ρa (11. σ= √ 2 λab T · V (11. 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 .38) (in which we can calculate −iMf i using Feynman diagrams).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. pn ).36) (11. dt (11.34) i. For the (proper) decay width one thus has 1 N Γ0 = .e. N 1 . 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. . . (11. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width 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 .35) This quantity is not covariant. . 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.T V.Scattering theory 108 Although this at ﬁrst sight does not look covariant.33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N.39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . Microscopically one has Ndecay = Na · Γ T or Γ= N 1 . . as expected. (11.

4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. q f ). θ). . 11.Scattering theory 109 (Using normalized wave packets these somewhat ill-deﬁned manipulations can be made more rigorous). (11. (11. (11. respectively.e. The result is N = |Mf i |2 dR(s. θ) in quantum mechanics.41) (11.43) dΩcm = π M (s. for which one has dσ/dΩ = |f (E. . . . see 11. (11.47) M (s. t) λab 4π dt. The sign diﬀerence comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . p1 . . . θ)|2 . 2 λab and for instance for two particles dσ = q ′ M (s. . 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. 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 .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. pn ). √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). pn ). . (11. . 16π 2 s (11. for which the √ amplitudes squared have been calculated in the previous chapter. . implies for the scattering matrix M . q n ). i. 8π s 2π 8π s . θcm ) √ q 8π s 2 2 (11.40) 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 . . p1 .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . p1 . pn ) |M |2 q 32π 2 m2 dΩ |M |2 . θ) or M (q i . q n ) q n qn √ Mni (q i . θ) = −8π s ℓ (in analogy with the expansion for f (E.42) 2−body decay The diﬀerential cross section (ﬁnal state not integrated over) is given by 1 dσ = √ |Mf i |2 dR(s.25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . ˆ ˆ Inserted in the unitarity condition for M .

(11.49) or Im Mℓ = q |Mℓ |2 . which allows writing Mℓ (s) = Sℓ (s) − 1 e2i δℓ (s) − 1 = . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni . 2i q 2i q (11. In general a given channel has |Sℓ (s)| ≤ 1. η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ . 0).e.51) and for the case that this is the only channel (purely elastic scattering. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. q ˆ Pℓ (ˆ · q ′ ) = q ˆ (ℓ) m 4π Y (ℓ) (ˆ) Ym (ˆ′ ) q (ℓ)∗ q 2ℓ + 1 m and the orthogonality of the Ym functions to prove that RHS = 2 n ℓ † (2ℓ + 1)Pℓ (ˆi · qf ) (Mℓ )f n qn (Mℓ )ni . the total cross section can be determined. ℓ (11.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 . θcm ) √ q 8π s 2 . (11. (11.53) The diﬀerence is the inelastic cross section. Using σ= dΩ q ′ M (s. (11.50) where Sℓ (s) satisﬁes |Sℓ (s)| = 1 and δℓ (s) is the phase shift. σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ). σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 .54) ℓ . Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ).48) If only one channel is present this simpliﬁes to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ . n (11. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)).52) From the imaginary part of M (s. q ˆ i.

This is (somewhat sloppy!) seen by considering the (amputated) 1PI two-point vertex Γ(2) = −i ∆ as the amplitude −iM for scattering a particle into itself (forward!) through the decay channels as intermediate states. This is achieved with a propagator i∆R (k) = i k2 − M2 + iM Γ (11. This is what we will do to discuss decay widths. however. such that |U (t. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . The quantity Γ is precisely the width for unstable particles.e. 0)|2 = e−Γt . 0) ∝ e−i(E−iΓ/2)t . ∝ U (t. i. For an unstable particle one expects that U (t. . but still real. 0).57) This shows that Γ is the width of the resonance. To check unitarity we need the amplitude at order e4 . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. (q Mℓ )ij (s) = −M Γi Γj . . which is given by a sum of the partial widths into the diﬀerent channels.Scattering theory 111 Note that the total cross section is maximal in the case of full absorption. the time-evolution operator. . 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 . 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.e. (11. σel = σinel . one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior.56) (again disregarding spin).55) (note that we have disregarded spin). The prescription for the pole structure. in which case. i. it is straightforward to write down the partial wave amplitude. We note that unitarity is generally broken in a ﬁnite order calculation. . s − M 2 + iM Γ (11. 11.58) .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. η = 0. For instance the tree-level calculation of Feynman diagrams at order e2 was real. The product of amplitudes is of order e4 . 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 .

59) reaching the unitarity limit for s = M 2 . The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb. u) + g(t. dt s(s − 4m2 ) with f (t. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. The cross section at resonance is about 30 nb. its width Γ = 120 MeV. u) + f (u. Its mass is M = 3096.4) can be written as 4πα2 dσ = {f (t. u) = g(t. From this one sees that a resonance has the same shape in all channels but diﬀerent strength. t) + g(u.1 Show that the cross section for electron-electron scattering (exercise 10.88 MeV. a fast change in the phase shift for a narrow resonance.Scattering theory 112 (Prove this using the unitarity condition for partial waves). Γ = 2. which is attributed to neutrinos. Knowing that each neutrino contributes about 160 MeV (see next chapter). The phase shift in the resonating channel near the resonance is given by tan δℓ (s) = MΓ . Three famous resonances are: • The ∆-resonance seen in pion-nucleon scattering. This characteristic shape of a resonance is called the Breit-Wigner shape. t)} . • The Z 0 resonance in e+ e− scattering with M = 91. 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). All these decays can be seen and leave an ’invisible’ width of 498 MeV. The half-width of the resonance is M Γ. Note that because of the presence of a background the phase shift at resonance may actually be shifted. the full width is Γ = 88 keV. This is a narrow resonance discovered in 1974. one can reconstruct the resonance shape for diﬀerent numbers of neutrino species. Limiting ourselves to a resonance in one channel. At resonance the cross section σT (π + p) is about 210 mb.26 keV. Three neutrinos explain the resonance shape. • The J/ψ resonance in e+ e− scattering. 2 q (s − M 2 )2 + M 2 Γ2 (11. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 2 1 1 ( s − m2 )( s − 3m2 ) tu 2 2 .2 GeV. Exercises Exercise 11. the partial width into e+ e− is Γee = 5.60) showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ.49 GeV. where furthermore σinel = 0.e. i. Its mass is M = 1232 MeV. M2 − s (11.

ρcm ρcm a b 1 cm . Calculate now fπ .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. we have here already used unitarity to ﬁnd a ﬁnite imaginary part that is proportional to the total decay width. Exercise 11. writing N (s) M1 (s) = 2 + iM Γ . a quantity which we will ¯ ¯ encounter in the next chapter (section 12. Thus. Compare the result with the factor trf 1/ρa ρb va that we used in the derivation of the basic expression for cross sections.8.999 877 Γ and Γ(π − → e− νe ) = 0.000 123 Γ.2).Scattering theory 113 Exercise 11. s − M∆ ∆ ∆ 2 calculate Re N (M∆ ) . cm |va − vb | in the center of mass system for two colliding particles a and b. 4.4 The ∆ resonance in N π scattering appears in just one partial wave (a P-wave) as a pole at the position 2 s = s0 = M∆ − iM∆ Γ∆ .6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Exercise 11. Compare the results with the experimental π − lifetime τ = 2.2 Calculate the ﬂux factor. Show that unitarity can be used one step further to ﬁx also the real part of the residue of the pole in the amplitude. As discussed for unstable particles.

Chapter 12

**The standard model
**

12.1 Non-abelian gauge theories

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

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

a

La

φ(x) = (1 + i θa La ) φ(x)

inf

(12.2)

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

inf

(12.3)

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

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

The standard model

115

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

′a a c Wµ = Wµ − cabc θb Wµ +

i (∂µ U (θ)) U −1 (θ), g

(12.7)

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

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

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

(12.8)

(12.9) (12.10)

The gauge-invariant lagrangian density is now constructed as

Gµν → U (θ) Gµν U −1 (θ).

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

(12.11)

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

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

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

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

**The standard model
**

xψ x+dx

116

(x)

ψ(x)

x+dx ψ(x+dx)

**x-space internal space
**

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

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

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

**A geometric picture of gauge theories
**

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

1 The

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

λ7 =

i

−i

1 1 λ8 = √ 3

−2

i. = . (12. (12. ds′ µ 0 ig Aµ (s) ψ(s) dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0).16) which in the presence of the ’connection’ Aµ diﬀers from the total change between x ψ(x) and x+dx ψ(x+ dx). x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x).18) or considering a path xµ (s) from a ﬁxed origin (0) to point x. ds s dxµ ds′ P exp ig A (s′ ) ψ(0). However. but expresses them with respect to diﬀerent bases.19) This gives rise to a (path dependent) phase in each point. 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. which connects two identical vectors. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0. -dx -dy µ µ dyµ µ dx For a vector. if two diﬀerent paths to the same point give diﬀerent phases the eﬀects can be observed. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. (12. 12. What is this physical eﬀect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.16.14) (12. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). The relation in Eq. ds which is solved by dψ(s) ds = dxµ .17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x).The standard model the basis for x G and x+dx 117 G. i. In principle such a phase in a given point is not observable. (12. A ’constant’ vector that only rotates because of the arbitrary deﬁnitions of local frames satisﬁes Dµ ψ(x) = 0. 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 .e.e. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x).15) ψ(x). should be independent of such transformations. Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x).

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

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

there must be a nonzero expectation value 0|Jµ (x)|π (k) . In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. π Example: chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD.. The generators Qa (in that case the rotation operators Lz . Taking the derivative. ψ −→ ei α·T ψ. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. a and a′ are degenerate states. Including a sum over the diﬀerent ﬂavors (up. denoted |π a (k) . T ] ψ. . ψu mu md ψ = ψd . down. H] = 0. T = τ /2. i. (12. etc.e.30) which in the case that we include only two ﬂavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. V µ = ψγ µ T ψ.The standard model 120 by degenerate representations of the symmetry group. The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).31) . M = . [Qa . 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. i. a massless Goldstone boson for each ’broken’ generator. Neglecting at this point the local color symmetry. a ∂ µ Jµ = 0. called ﬂavors. ∂ ∂ (12. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . and one must have mπa = 0. Goldstone mode. (12. it is easy to see that one gets ∂µ V µ = i ψ [M. A bit more formal.e. if the generators Qa generate a symmetry. irrespective of the fact if they annihilate the vacuum. (12. Using the Dirac equation. π (12.26) for all the states labeled by a corresponding to ’broken’ generators. strange. . = ψ(i/ − M )ψ. i.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved. Using translation invariance and as kµ is the only four vector on which this matrix element could depend one may write a 0|Jµ (x)|π b (k) = fπ kµ ei k·x δab (fπ = 0) (12.29) (Note that each of the entries in the vector for ψ in the above equation is still a 4-component Dirac spinor). .28) where ψ is extended to a vector in ﬂavor space and M is a diagonal matrix in ﬂavor space. This means that they are operators that create states from the vacuum.e. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). .) one can write L = f ψ f (i/ − Mf )ψf . suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). 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. This lagrangian density then is invariant under unitary (vector) transformations in the ﬂavor space. 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′ .

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. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. (12. chiral symmetry is not perfect. From the number of broken generators it is clear that one expects three massless Goldstone bosons. the lagrangian density is invariant under left (L) and right (R) rotations independently. How can we see this. ψ −→ ei α·T γ5 ψ. generated by TR/L = 1 (1 ± γ5 )T .27. I. or. T ] = 0. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions.The standard model 121 Furthermore ∂µ V µ = 0 ⇐⇒ [M.33) which in the case of two ﬂavors form SU (2)A transformations generated by the Pauli matrices.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. while the groundstate is only invariant under isospin rotations (R = L). parity would not play a role in the world. The chiral ﬁelds ψR and ψL are transformed into each other under parity. Note that these transformations also work on the spinor indices. 12. fπ = 0 remains. i. T } γ5 ψ. etc.35 is nonzero. (12.34) Using the Dirac equation. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry.e.35) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. i. . which is approximately true for the up and down quarks. socalled axial vector transformations. where the quark masses are not completely zero. Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. for which the ﬁeld (according to the discussion above) has the same behavior under parity. T γ5 = τ γ5 /2. From group theory (Schur’s theorem) one knows that the latter can only be true. both are very small. it is easy to see that one gets ∂µ Aµ = i ψ {M. ψR/L = 1 (1 ± γ5 )ψ. 12.3 MeV/c2 and Mn = 939. proton and neutron. In the real world. as the quantity ∂µ Aµ (x). The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . behaves as a pseudoscalar particle (spin zero. e. but also a triplet (isospin 1) of pions. M = m · 1. This situation. SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. There exists another set of symmetry transformations. Therefore the world of up and down quarks describing pions. ∂ ∂ (12. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. in terms of which the Dirac lagrangian 2 density looks like L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR . i.e.e. (leaving out the ﬂavor structure) the same as ψγ5 ψ. is by nature not realized in the Weyl mode. which is equivalent 2 to the V-A symmetry. however. a doublet (isospin 1/2) of nucleons. nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . including in Dirac space a γ5 matrix. much smaller than any of the other mesons or baryons. etc.e. gives also rise to a nonzero mass for the Goldstone bosons according to Eq. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A .6 MeV/c2 ). but the fact that the symmetry is not perfect and the right hand side of Eq.e. This combined symmetry is what one calls chiral symmetry. Aµ = ψγ µ T γ5 ψ. stated equivalently. with almost degenerate masses (Mp = 938. is what happens in the real world. if in ﬂavor space M is proportional to the unit matrix. This symmetry.g. parity minus). (12. We know that mesons and baryons (such as the nucleons) have a well-deﬁned parity that is conserved.

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

implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. 1 As they are massless. left. explicitly L R SU(2)W SU(2)W −→ ei α·τ /2 L. ψR/L = 2 (1 ± γ5 )ψ decouple. Thus the lagrangian density is L (f ) ∂ ∂ = i L/ L + i R/ R. ∂ ∂ ∂ (12. explicitly L −→ e−i β/2 L.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. νµ and ντ ) and their antiparticles. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L. (12. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ.50) (12. The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . (12. R −→ e−i β R. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature.51) . 2 2 = ∂µ R + i g ′ Bµ R. etc.and righthanded particles.47) and YW is considered as an operator that generates a U (1)Y symmetry. = ∂µ L − (12.49) Next the SU (2)W ⊗ U (1)Y symmetry is made into a local symmetry introducing gauge ﬁelds Wµ and Bµ in the covariant derivative Dµ = ∂µ − i g Wµ · TW − i g ′ Bµ YW /2.). muons. R. τ ). For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. etc.The standard model 123 12. under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 . The procedure.e. By introducing gauge ﬁelds and breaking the symmetry a more complex lagrangian is obtained. 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. neutrinos.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. while the lefthanded particles form a doublet. This is written as 3 Q = TW + YW .45) (12. the quarks (up. i. down. 2 (12. muon µ− or tau τ − ). however. its corresponding neutrino (νe .48) (12.44) which has an SU (2)W symmetry under transformations eiα·TW . the leptons (elektrons. electromagnetic and weak forces. U(1)Y U(1)Y (12. that gives a good description of the physical world.43) and One introduces a (weak) SU (2)W symmetry under which eR forms a singlet.) and the gauge bosons responsible for the strong.

with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. . h) (12. the U (1)Q symmetry (generated by the charge operator). . L where L (h1) (h) =L (h1) +L (h2) . † i i g Wµ · τ − g ′ Bµ )φ. . (12. TW = ±1 or 0 and YW = 0 (thus Q = TW ) and 3 Bµ is a singlet under SU (2)W (TW = TW = 0) and also has YW = 0. (12.59) . which assures with the choice of YW for the Higgs ﬁeld that Q generates the remaining U (1) symmetry. (12. . 2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). + 8 g2 v2 1 + − (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + . + 8 (12. . a complex Higgs ﬁeld + 1 φ √2 (θ2 + iθ1 ) 0 = 1 (12. 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. 4 4 In order to break the symmetry to the symmetry of the physical world.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 . Using local gauge invariance θi for i = 1.58) (12. −V (φ) L and (h2) = −Ge (LφR + Rφ L). this leads to the lagrangian L (h1) .54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 .56) v + h(x) 2 Dµ φ = (∂µ − and Dµ φ = = 1 √ 2 − ig 2 ∂µ h + 1 2 Wµ −iWµ √ (v + h) 2 3 gWµ −g′ Bµ i √ (v + 2 2 Up to cubic terms.53) φ= √ (θ4 − iθ3 ) φ 2 with TW = 1/2 and YW = 1 is introduced.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 gWµ − g ′ Bµ + . Putting this in leads to L (f ) =L (f 1) +L (f 2) . For the classical ﬁeld the choice θ4 = v is made.The standard model 124 3 3 where Wµ is a triplet of gauge bosons with TW = 1.

g ′ /g = tan θW . 2 Zµ (12. . we can read oﬀ e = g sin θW = g ′ cos θW . 2 + g ′2 g 3 g ′ Wµ + g Bµ (12.60) (12. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12. (12. (12. 2 (12. The coupling of the fermions to the physical gauge bosons are contained in L (f 1) giving L (f 1) = i eγ µ ∂µ e + i νe γ µ ∂µ νe − g sin θW eγ µ e Aµ g 1 1 + sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe cos θW 2 2 g µ + µ − + √ νe γ eL Wµ + eL γ νe Wµ . . 4 2 MW g2 v2 = .69) .66) From the coupling to the photon.63) (12.62) g 2 + g ′2 3 ≡ sin θW Wµ + cos θW Bµ .64) (12.65) The weak mixing angle is related to the ratio of coupling constants. 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 .67) The coupling of electrons or muons to their respective neutrinos. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .61) (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 .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 . 2θ 2θ 4 cos W cos W 0. for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + .

3 = TW . Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed.The standard model 126 the good old four-point interaction introduced by Fermi to explain the weak interactions. 1. −5 (12. one obtains L (h2) Ge v Ge = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me = −me ee − eeh.05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron.74) (12. Comparing this with the total width into (invisible!) channels. me /v is extremely small. .19 GeV. (12.71) GeV −2 = = = = . Γe ≈ Γµ ≈ Γτ ≈ 83 MeV).78) From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 80. Finally we want to say something about the weak properties of the quarks.70) = 2 = 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.77) (12.72) (12. g ′ and v determine a number of experimentally measurable quantities.5 MeV (exp.80) First. (12. 2 The value of α deviates from the known value α ≈ 1/137 because of higher order contributions. With the choosen vacuum expectation value for the Higgs ﬁeld. 91. muon or tau. from the MW mass is about 250 GeV.166 4 × 10 0.73) (12.e. 2 2 2 (12.76) with CV CA 3 = TW − 2 sin2 θW Q. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.e.231 2.40 GeV. (12. g2 MZ 2 2 (CV + CA ). one has the relation g2 e2 1 GF √ = . i.g. giving rise to a running coupling constant after renormalization of the ﬁeld theory. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). (12. The coupling to the Higgs particle is weak as the value for v calculated e. i. the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).75) The coupling of the Z 0 to fermions is given by (g/ cos θW )γ µ multiplied with 3 TW 1 1 1 (1 − γ5 ) − sin2 θW Q ≡ CV − CA γ5 . such as2 α = e2 /4π GF sin2 θW MW MZ ≈ 1/127. v (12. The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . 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.

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

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. The Higgs ﬁeld is still limited to one complex doublet. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR .q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. ms and mb ).85) √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu . φ = −φ− 0 2 For the (squared) complex matrices we can ﬁnd positive eigenvalues.g.83) where Vu and Vd are unitary matrices. † Λu Λ† = Vu G2 Vu . mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . each of these containing the three families. i.22 and η ≈ 0. d d (12. (12. the weak mass eigenstates are L (h2. Note that we need the conjugate Higgs ﬁeld to get a U (1)Y singlet in the case of the charge +2/3 quarks. e. UR = uR cR tR . Thus one has † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. A ≈ 0. respectively).q) ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL .86) the unitary CKM-matrix introduced above in an ad hoc way. (12.The standard model 128 using the socalled Wolfenstein + O(λ4 ) with λ ≈ 0.227. 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 . Working with the mass eigenstates one simply sees that the weak current coupling to the W ± mass † mass becomes U L γ µ DL = U L γ µ Vu Vd DL . The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. . The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† .e. for which we need the appropriate weak isospin doublet 0∗ 1 φ c = √ v +h . u u where we include now the three lefthanded quark doublets in QL . allowing us to write † Λu = Vu Gu Wu † and Λd = Vd Gd Wd .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 (h2.34.82) space.82 and ρ ≈ 0.84) with Gu and Gd being real and positive and Wu and Wd being diﬀerent unitary matrices.

88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet.5×10−3 eV2 .90) mass † mass mass † mass † We note that there are mass ﬁelds ER = We ER . For neutrino’s this matrix is parametrized in terms of three angles θij with cij = cos θij and sij = sin θij and one angle δ. The 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 . For the (massless) neutrino ﬁelds we just can redeﬁne ﬁelds into NL = Ve† NL . EL = Ve EL . while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2.e.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 . The mass ﬁelds ER † mass † mass = We ER . we ﬁnd massless neutrinos. As before. (12. e N L= L EL (12. mµ and mτ . it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). 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 . † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . since the weak current is the only place where they show up. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . there is no family mixing for massless neutrinos. and obtain L (h2. one can write Λe = Ve Ge We and we ﬁnd † L (h2.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL .ℓ) = −LφΛe ER − ER Λ† φ† L. UPMNS a parametrization that in principle also could have been used for quarks. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL . (12.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix. (12. = − 1/6 2/3 0 1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. decoupling from all known interactions.ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . i. In this case. EL = Ve† EL .The standard model 129 The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2.ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12. (12.89) with three righthanded neutrinos added to the previous case. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2.92) .

2 although this option is not attractive as it violates the electroweak symmetry.94) M L NL NL + M L NL NL . the left. 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. The way to circumvent this is to introduce as in the previous section righthanded neutrinos. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass. The see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. For the neutrino sector. couple the right.ν = − MR NR NR + MR NR NR .95) 2 In order to have more than a completely decoupled sector.96) VPMNS = UPMNS K with K = 0 eiα2 /2 0 .and righthanded species then just form a Dirac fermion. 0 0 1 L mass. (12. are equivalent to two decoupled Majorana neutrinos (see below).99) . This is called the see-saw mechanism (outlined below). − MD MR nR 2 (12. one must for the neutrinos as well as charged leptons. For the leptons. The mass matrix can be written as ML |MD | eiφ M = (12. We use here the primes starting with the weak eigenstates. Actually. hence the mixing matrix becomes iα /2 0 0 e 1 (12. is to add in Eq. one being massive. however.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. with for the righthanded sector a mass term MR .c.ν =− containing three (CP-violating) phases (α1 .The standard model 130 The lepton sector (massive Majorana ﬁelds) An even simpler option than sterile righthanded Dirac neutrinos. α2 and δ). 2 R ψ = nR + nL . Thus 2 Υ ′ = nc + nL . as above.100) |MD | eiφ MR 1 1 ∗ ∗ M L nc nL + M L nL nc − M R nR nc + M R nc nR L R L R 2 2 1 1 ∗ ∗ − M D nc nc + M D nL nR − M D nR nL + M D nc nc L R R L 2 2 1 nc nR M L M D nL L c + h. (12. the massless and massive Majorana neutrinos. 1 L Υ ′ = nR + nc . 12. Υc ≡ (ΥL )c = ΥR 1 2 L and Υc = ΥL ). For these light Majorana neutrinos one has.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.98) (12. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. Thus (disregarding family structure) one has two Majorana neutrinos. Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6. Absorption of phases in the states is not possible for Majorana neutrinos. 1 ∗ c c (12. 1 c ∗ c L mass.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. a unitary matrix relating them to the weak eigenstates. coupled by a Dirac mass term.

For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 ﬁnds M1 ≈ MD /MR and M2 ≈ MR . H] = 0. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. a and a′ are degenerate states. Exercises Exercise 12. T ] ψ. Υ2R nR (12. Exercise 12.2 Derive for the vector and axial vector currents. . (12.The standard model 131 taking ML and MR real and non-negative.103) for each of which one ﬁnds the lagrangians L = and we are left with two decoupled Majorana ﬁelds Υ1L = U ∗ nc . (12. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. 2 with real masses Mi .101) Thus one obtains from 2 ML + |MD |2 † MM = |MD | ML e+iφ + MR e−iφ the eigenvalues 2 M1/2 |MD | ML e−iφ + MR e+iφ 2 MR + |MD |2 . (12.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. i ψ {M. T } γ5 ψ. [Qa . Exercise 12.e. 2 U (M M † ) U † = M0 . The diagonalization is analogous to what was done for the Λ-matrices and one ﬁnds U M U T = M0 with a (unitary) matrix U . which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † .105) for i = 1. Prove that if the generators Qa generate a symmetry. 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).e. i.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i g 2 cos θW f f CV γ µ − CA γ µ γ5 . i. 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′ .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)) . L Υ2L nR Υ1 and Υ2 . related via c Υ1R = U nL .1 Consider the case of the Weyl mode for symmetries.

Γ(Z 0 → νe νe ).7 In this exercise two limits are investigated for the two-Majorana case. hhh. ghbb = b v v where v is the vacuum expectation value of the Higgs ﬁeld. Mf M2 and ghf f = .The standard model 132 with CV CA = = 3 TW − 2Q sin2 θW . In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . to be precise one has for bosons (b) or fermions (f). . Exercise 12. Exercise 12.231.) are proportional to the mass squared for bosons or mass for fermions of the particles. and the width to a pair of neutrino’s. Exercise 12. at least up to a high (which?) order in λ. The mass of the Z 0 is MZ = 91 GeV. 0 ¯ Γ(Z 0 → f f ) = MZ g2 48π cos2 θW f f CV 2 + CA 2 . Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν . Write down the matrix element squared (averaged over initial spins and summed over ﬁnal spins) for the decay of the Z 0 . etc.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix.5 Show that the couplings to the Higgs (for W + W − h. 2 2 The masses are mt ≈ 175 GeV. ZZh. (b) Calculate the width to electron-positron pair.4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. Note that you can ﬁnd the answer without explicit construction of all interaction terms in the lagrangian. the weak mixing angle is given by ¯ sin2 θW = 0. mb ≈ 5 GeV and MW ≈ 80 GeV. e+ e− h. 3 TW . ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). Exercise 12. Γ(Z 0 → e+ e− ).

Given that neutrino masses are of the order of 0.05 eV. what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV). . which leads to the socalled see-saw mechanism. Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 .The standard model 133 (a) Calculate for the special choice ML = MR = 0 and MD real. the mass eigenvalues and show that the mixing matrix is 1 1 1 √ U= i −i 2 which enables one to rewrite the Dirac ﬁeld in terms of Majorana spinors. Calculate the eigenvalues ML = 0 and MR = MX . (b) A more interesting situation is 0 = ML < |MD | ≪ MR .

- An Introduction to Quantum Field Theory - Peskin
- Lahiri and Pal-A-First Book of Quantum Field Theory-2nd Ed
- Weinberg, Steven - The Quantum Theory of Fields Volume II
- An Introduction to Quantum Field Theory(Peskin)
- HW1
- Srednicki M. Quantum Field Theory - Solutions Manual(Web Draft, 2006)(156s)_PQtb
- Elementary Quantum Field Theory
- Huang K. - Quantum Field Theory
- quantum field theory
- Srednicki Ms Qft Solutions Rev
- Kleinert H. Path Integrals in Quantum Mechanics, Statistics, Polymer Physics, And Financial Marke
- Introduction to quantum field theory, based on the book by Peskin and Schroeder
- Weinberg the Quantum Theory of Fields Vol 1 Foundations
- PESKIN - Quantum Field Theory
- Quantum Field Theory 2013
- Beata J. Dąbrowska-Wüster et al- Multicomponent gap solitons in spinor Bose-Einstein condensates
- MolecularThermodynamics Quantum Part Atkins9e
- Fundamentals and Dirac Formalism in QM (G. Ingelman)
- Grace's Comps
- The Fourier Grid Hamilton Ian Method for Bound State Eigenvalues
- Autotoformas e Quantum
- PhysChem 1 Worksheet 002
- Cuadrivectores en Electromagnetismo
- Solutions to Levi Applied Quantum Mechanics 2nd Ed
- Uncertainty Principle for Dummies
- JEST Physics Full Solved Sample Paper
- Modeling and Control of Offshore Pipelay Operations Based on a Finite Strain Pipe Model
- Handout09 a2 Useful
- Fossen
- Pages From Lecture2013

Sign up to vote on this title

UsefulNot usefulClose Dialog## Are you sure?

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

Loading