QUANTUM FIELD THEORY

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

September 2009 (version 4.1)

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 17 19 20 26 26 28 30 33 35 38 38 38 40 43 43 45 47 48 49 53 53 55 56 59

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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 61 65 65 66 67 68 69 72 72 74 77 78 80 81 84 84 88 91 94 99 102 102 105 106 108 110 113 113 117 121 122 126

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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

Cambridge University Press. Vol.3 2. 1995.2 3.3 2.2 6.3 3. Peskin and D. Itzykson and J.2.1 3. These notes Section 1. 2.E.2 4. Besides the books.3 2.2 2.3 3.2 2. Peskin & Schroeder and Srednicki. 4. Cambridge University Press. Zuber. II: Modern Applications. An introduction to Quantum Field Theory.2 6. Veltman.2 .-B. McGraw-Hill.3 3.3 2. 1996. L. e. The schedule in the Fall of 2009 is Chapters 1 through 6 in period 1 (7 weeks in September and October 2009). Corresponding chapters in books of Ryder.g.3 3. 6.4 2. Addison-Wesly. 1994.5 5.4 4. Ryder. The quantum theory of fields.3 2. M.2 2. Quantum Field Theory.3 References As most directly related books to these notes. 3.1 2. 5. Quantum Field Theory.7 2.3 6.3 3. 1. Cambridge University Press. M.2 3. there is also material available via the web pages of colleagues.2 3. 3.1 6.8 3.4 2. Schroeder.1 3.2 3.3. Srednicki. Chapters 7 through 10 in period 2 (7 weeks in November and December 2009) and Chapters 11 and 12 in period 3 (3 weeks in January 2010). Other text books of Quantum Field Theory that are useful are given in refs [3-6].H. 1995. M. I refer to the book of Srednicki [1] and Ryder [2].4 6.V. The notes contain all essential information. Vol. S. 2007. but are rather compact.1 1.1 2 2 3.5 2.4 3.3 4. I: Foundations.4 3. Weinberg. C. Diagrammatica.2 1.1 5.1 4. Quantum Field Theory. those of Jan Smit (UvA).5 4.5 2.4 3. Pierre van Baal (UL). 1985. Bert Schellekens (RU and NIKHEF) and Gerard ’t Hooft (UU). 3. Cambridge University Press.1.1 2.5 Ryder Peskin & Schroeder Srednicki 1 1 1 1 3 2. 2. 1980.2 3. Cambridge University Press.

4 10.2 8.2 11. 9.3 8.5 6.2.1.3 8.4 These notes Section 7.3 20. 4.1.4 4.5 6 7 6 6.10 8.5 Ryder 4 4 4 4 4 4 Peskin & Schroeder 2.2 7.1 20.7 4.4 12.7 5.5 4.3 9.2 12.1 5.5 7.6 4.5 11.5 2.4 7. 6.2 4.6 3.1 12.2 10. 4.1 10.3 6 8 9 Srednicki 3 3 3 5.4 8.1 11.2 5. 6.3 10. 4.4.4.6 3. 6.3. 20. 2.4 3.8 5 4.3.4.6 3.2.2.5.3.1 20.4 12.5 9. 6.3 7.2 9.1 9.6 10.7. 4.2 8.1.4 2. 6.5 4. 8.2.3 3.3 7.3 2.1 7.3 11.3 6.6 6.8 6. 9.6 8.6 9. 2.5 7.1 8.1 9.5 9.1 20.3 12.5 .4 9.

5 .

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

The set of such amplitudes p′ . we find that the original particle cannot be located better than within a distance ¯ /mc. p2 .Introduction 2 and ψ(r) = d3 p exp (2π¯ )3 h i ˜ p · r ψ(p). . For a particle in interaction. Let us consider the measurement of the position of a particle. The only case in which the momentum of a particle can be measured exactly is when the particle is free and stable against decay. ) of free particles. an atom is a perfect example of a particle. however.2 Units It is important to choose an appropriate set of units when one considers a specific problem. . For a massless particle one gets ∆x ≫ ¯ /p = λ/2π. i. which have been discussed in the previous section. Therefore if h we want ∆x ≤ ¯ /mc (where m is the rest mass of the particle). For the momentum or energy of a particle we know that in a finite time ∆t. the better name might be ’degree of freedom’. These are the particles in the initial and final state of a scattering process. In this case the momentum is conserved and one can let ∆t become infinitely large. In fact. h ∆x ≥ h ¯ . mc (1.7) The existence of a limiting velocity. The theory will not give an observable meaning to the time dependence of interaction processes. out|p1 . This implies that the momenta of particles can only be measured exactly h when one has an infinite time available. 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. 2 1 1 2 Another point that needs to be emphasized is the meaning of particle in the above context. the coordinates of a photon only become meaningful in cases where h the typical dimensions are much larger than the wavelength. its Compton wavelength. 1. p2 . 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 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. it is the behavior under Poincar´ transformations or in the limit v ≪ c Gallilei transformations that determine the description e of a particle state. p′ . in particular the free particle state. If the particle momentum becomes relativistic. Actually. h ¯ (1. It is then possible to create a particle . momenta of the order p ∝ mc and h energies of the order E ∝ mc2 are involved.antiparticle pair and it is no longer clear of which particle we are measuring the position. the momentum changes with time and a measurement over a long time interval is meaningless. This is true specifically for the domain of atomic. If the energy is low enough to avoid excitation of internal degrees of freedom. where the typical numbers are difficult to conceive on a macroscopic scale. Any device that wants to locate the position of say a particle within an interval ∆x will contain momentum components p ∝ ¯ /∆x. in ≡ p′ . Some consequences are that only in cases where we restrict ourselves to distances ≫ ¯ /mc. the h concept of a wave function becomes a meaningful (albeit approximate) concept. which says that h a particle cannot be located better than its de Broglie wavelength. As a result. in|S|p1 . . nuclear and high energy physics. the energy uncertainty is given by ∆E ≥ ¯ /∆t. one has E ≈ pc and ∆x ≥ ¯ /p.8) For a moving particle mc2 → E (or by considering the Lorentz contraction of length) one has ∆x ≥ h ¯ c/E.e. They are governed by a few fundamental units and constants. leads to new fundamental limitations on the possible measurements of physical quantities. p′ . because physical sizes and magnitudes only acquire a meaning when they are considered in relation to each other. in determines the scattering matrix or S-matrix. Thus the coordinates of a particle cannot act as dynamical variables (since these must have a precise meaning). namely h and ¯ . This position cannot be measured with infinite precision.

g. In some cases like the electromagnetic and strong interactions. given a length l the quantity l/c has the dimension of time and one finds 1 m = 0. e.672 621 71 (29) × 10 −27 2 (1. one needs h ¯ 2 c2 = 0.11) (1. ¯ All length.13) kg = 0.14) . time and energy or mass units then can be expressed in one unit and powers thereof. it turns out to be convenient to work with units such that h and c are set to one. 4π ǫ0 ¯ c h (1. (1.035 999 11 (46).1).9) or for order of magnitude estimates hc ≈ 0.33 × 10−8 s or eliminating the second one would use that. is eliminated. e.109 382 6 (16) × 10−31 kg = 0. the quantity ct has dimension of length and hence 1 s = 3 × 108 m. cross sections.g. We could as well have set c = 1. for which one can use energy (see table 1. given a time t.Introduction 3 c. me mp = 9. meter or second. The elementary unit that is most relevant depends on the domain of applications. 2 Furthermore one encounters the strength of the various interactions.g.g.12) (1. e. we can work with less units. To convert to other units of length or time we use appropriate combinations of h and c. e.2 GeV fm = 200 eV nm.938 272 029 (80) GeV/c .389 379 323 (67) GeV2 mbarn (1 barn = 10−28 m2 = 102 fm2 ). For areas. 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/¯ h l/(¯ c) h E pc ℓ/¯ h mc2 A/(¯ c)2 h F ¯c h α = e2 /4πǫ0 ¯ c h GN /(¯ c5 ) h 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.510 998 918 (44) MeV/c .197 326 968 (17) GeV fm (1.1: Physical quantities and their canonical dimensions d. Depending on the specific situation. For times one needs h ¯ = 6. the eV for atomic physics.g.5 × 1024 GeV−1 . the quantity c is used to define the meter.582 119 15 (56) × 10−22 MeV s.g. for electromagnetism the fine structure constant α= e2 = 1/137. = 1. of ¯ course masses come in that one needs to know or look up. e. implying (when h = c = 1) that ¯ ¯ 1 fm = 10−15 m ≈ 5 GeV−1 . By making use of these fundamental constants.10) implying (when h = c = 1) that 1 s ≈ 1. determining units (energy)d . the MeV or GeV for nuclear physics and the GeV or TeV for high energy physics. For instance. This would mean that one of the two units. e. those of the electron or proton. Table 1. these can be written as dimensionless quantities. for lengths ¯ h ¯ c = 0.

1. A vector x can be expanded with respect to a basis ei (i = 1. It remains useful. Useful relations are ǫijk ǫimn ǫijk ǫijl = = δjm δkn − δjn δkm . ˆ x= i=1 xi ei = xi ei . The inner product of two vectors is a rank 0 tensor or scalar. Having many particles. for gravity Newton’s constant. one can also eliminate the last dimension.. (1. i linear transformations that do not change the length of a vector. ˆ ˆ (1. All masses. 2. a∞ = 4πǫ0 ¯ 2 /me e2 or the Bohr magneton µe = e¯ /2me . the metric in E(3) is given by ei · ej = δij .16) Quantities that do not contain h or c are classical quantities. GN = 6. xi . A vector is a tensor of rank 1.. to use h λ ¯ ¯ and c to simplify the calculation of quantities. lengths and energies are compared with the Planck mass or length (see exercises). . The Kronecker delta is a constant rank-2 tensor. 3 or i = x. Tensors of rank n are objects with n components that transform according to Ti′1 . Rin jn Tj1 . e. in which case zi = ǫijk xj yk . summation over this index is assumed (Einstein summation convention).in = Ri1 j1 .22) .19) The inner product of a vector with itself gives its length squared. A relation with energy is established via the average energy of a particle being of the order of kT . the electron Compton wavelength −e = h/me c. e. When a repeated index appears. however. 2 δkl . (1.17) to get the three components of a vector.g..g. y. (1. the Bohr radius. e. the electron rest energy me c2 and the classical electron radius re = e2 /4πǫ0 me c2 . Quantities that only h h contain c occur in classical relativity. (1.21) (1. ˆ ˆ (1.18) 0 if i = j. such as on the right hand side of this equation.20) ǫijk =  0 otherwise. rotations are those real. where the Kronecker delta ˆ ˆ is given by 1 if i = j δij = .380 650 5(24) × 10−23 J/K = 8. 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 (1. ¯ e. .3 Conventions for vectors and tensors 3 We start with vectors in Euclidean 3-space E(3). Quantities that contain both h and c play a role in relativistic quantum ¯ mechanics. that can be used in the cross product of two vectors z = x × y.g. with the Boltzmann constant given by k = 1.708 7 (10) × 10−39 GeV−2 . The inner product of two vectors is given by x · y = xi yi ei · ej = xi yj δij = xi yi . Actually. Choosing an orthonormal basis. x′ = Rx or x′ = Rij xj leading to a new vector with different components.g. It is an invariant tensor that does not change under rotations.Introduction 4 For weak interactions and gravity one has quantities with a dimension. the mass of the electron me . z). e.g.jn .617 343(15) × 10−5 eV/K. ¯ Quantities that contain only h are expected to play a role in non-relativistic quantum mechanics. A vector can be rotated..15) h ¯ c5 By putting this quantity equal to 1. the concept of temperature becomes relevant.

. such as eigentime τ for the position vector τ 2 ≡ x2 = t2 − x2 .2. The (invariant) lengths often have special names. .Introduction 5 We note that for Euclidean spaces (with a positive definite metric) vectors and tensors there is only one type of indices. p) = (E. we will use upper indices for the three-vectors.26) .νn can be conν structed. is timelike with invariant length (squared) p2 = p · p = pµ pµ = E 2 − p2 = m2 . So we could have used all upper indices in the above equations. The invariant distance between two points x and y in Minkowski space is determined from the length dsµ = (x − y)µ . linear transformations that do not change the length of a four-vector are called the Lorentz transformations. ˆ ˆ (1. xµ = gµν xν . . ∂t (1. x2 .25) (1. ∂t ∂x ∂y ∂z = ∂ . ∂3 ) = ∂ ∂ ∂ ∂ .. (1. • ds2 = 0 (lightlike or null intervals). These transformations do change the components of a vector. Relations relating tensor expressions. x2 .∇ . For four-vectors in Minkowski space we will use the notation with upper indices and write x = (t. The Poincar´ transformations include e e Lorentz transformations and translations. The real. x2 = x · x = xµ xν nµ · nν = xµ xν gµν . 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. are called covariant. the invariant length or distance (squared) is not positive definite.µn = Λµ1ν1 . linear transformations that leave the length of a vector invariant are called (homogeneous) Lorentz transformations. it is convenient to distinguish lower indices from upper indices. x1 .23) The quantity gµν ≡ nµ · nν is the metric tensor. 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 . The lower indices are constructed in the following way. x1 . independent of a coordinate system. In Minkowski space. Vectors are denoted x = xµ nµ . where m is called the mass of the system. When 3-dimensional space is considered as part of Minkowski space. One has x2 = xµ xµ = t2 − x2 . Λµn n T ν1 . Note that in this equation one has on left. the points lie on the lightcone and they can be connected by a light signal. ∂2 . and are given by (x0 . x3 ). Because of the different signs occurring in gµν .1. xi are the three space components. • ds2 < 0 (spacelike intervals). No difference is made between upper or lower.3) with basis nµ (µ = ˆ 0. The length (squared) of a vector is obtained from the scalar ˆ product.. denoted as V ′µ = Λµν V ν . Unlike in Euclidean space.. 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 . ∂1 .. The derivative ∂µ is defined ∂µ = ∂/∂xµ and we have a four vector ∂ with components (∂0 . given by g00 = −g11 = −g22 = −g33 = 1 (the ˆ ˆ other components are zero). x3 ) = (t. where the coordinate t = x0 is referred to as the time component. x) = (x0 . one must distinguish tensors with upper or lower indices and one can have mixed tensors. Many other four vectors and tensors transforming like T ′ µ1 . −x).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. . Examples are the scalar products above but also relations like pµ = m dxµ /dτ for the momentum four vector.3). however. 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.24) Within Minkowski space the real. explicitly written as (p0 . p). In special relativity we start with a four-dimensional real vector space E(1. One can distinguish: • ds2 > 0 (timelike intervals).

(c) Calculate the Compton wavelength of the electron and the quantities under (a) and (b) using the value of hc.27) ν ν Note that gµ with one upper and lower index. ǫ0 .28) 2 = ∂ µ ∂µ = ∂t2 The value of the antisymmetric tensor ǫµνρσ is determined in the same way as for ǫijk . (1.32) (1.30) ǫµνρσ ǫµν ′ ′ ρσ ′ = − (1. For the contraction of a vector with the antisymmetric tensor one often uses the shorthand notation ǫABCD = ǫµνρσ Aµ Bν Cρ Dσ . The length squared of ∂ is the d’Alembertian operator. α and me c2 = 0. c.Introduction 6 It is easy to convince oneself of the nature of the indices in the above equation.31) . Relate this quantity to the electron Compton wavelength −e via the dimensionless h λ fine structure constant α. gµ = gµρ g ρν and 0 1 2 3 is in essence a ’Kronecker delta’. a∞ = 4πǫ0 ¯ 2 /me e2 .1 (a) In the the Hydrogen atom (quantum system) the scale is set by the Bohr radius. (1. ′ ′ ǫ µνρσ ǫµνρσ The first identity. (1. re = e2 /4πǫ0 me c2 to the Compton wavelength. (1.29) (Note that there are different conventions around and sometimes the opposite sign is used). is easily proven for ǫ0123 ǫ0123 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. but only appropriate combinations.33) (1.511 MeV. (1. defined by ∂2 − ∇2 . for instance. but more important has the same effect on lefthandside and righthandside. e. Each of these permutations leads to a minus sign. This demonstrates how a careful use of units can save ¯ a lot of work. (b) Relate the classical radius of the electron (a relativistic concept). = −24. It is an invariant tensor. because one has ν ∂µ xν = gµ . ′ ′ . 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 . ¯ . constructed via the metric tensor itself. me . starting from ǫ0123 = 1.34) ρ ǫµνρσ ǫµν σ σ ǫµνρσ ǫµνρ ′ ′ = −2 g ρρ g σσ − g ρσ g σρ = −6g σσ . One does not need to know h. not affected by Lorentz transformations. g0 = g1 = g2 = g3 = 1. (1.35) Exercises Exercise 1.

ˆ ˆµ n3 = (0. . a3 ) are the expansion coefficients using the basis vectors n0 .13 to give its actual value in kg.71 × 10−39 GeV−2 to construct a mass h Mpl (Planck mass). n1 . n3 . ν.3. ρ. 0. 0. a2 ] or [a− . 0. ]. 1. a+ . 0. 1)/ 2 = (ˆ 0 + n3 )/ 2 ˆ n ˆ √ √ n− = (1. 3} the index α can only be equal to one of the indices in ǫµνρσ . h h [Note: what is the MKS unit for V/T?] Exercise 1. a1 . 2. a1 . which implies ˆ µ that the components of nα are given by nα = gα . n2 . a1 . a2 . 0. by a simple few-line reasoning [For instance: If {µ. 1. g−− . which is the Compton wavelength for the Planck mass.(A · B) C using the properties of the tensor ǫijk given in section 1.3 Prove the following relation ǫµνρσ g αβ = ǫανρσ g µβ + ǫµαρσ g νβ + ǫµνασ g ρβ + ǫµνρα g σβ . n2 .4 Lightcone coordinates for a four vector a (which we will denote with square brackets as [a− . n1 . a+ . (a) Express the scalar product a · b in lightcone coordinates and deduce from this the values of g++ . (b) The coordinates (a0 . g+− and g−+ . σ} is a permutation of {0.2 Prove the identity A × (B × C) = (A · C) B . Exercise 1. Compare it with the proton mass and use Eq. 0. a+ . n+ . 0. where ˆ ˆ ˆ ˆ n1 = (0. . 0) ˆ n0 = (1. −1)/ 2 = (ˆ 0 − n3 )/ 2 ˆ n ˆ Show that the components of a four vector can be obtained through a · nα = aα . 0. . 1. aT ]) are defined through √ a± ≡ (a0 ± a3 )/ 2. 1.Introduction 7 (d) Use the value of the gravitational constant GN /¯ c5 = 6. 0. 1) ˆ . Also construct and calculate the Planck length Lpl . 0. 0) ˆ √ √ n+ = (1. Exercise 1. ˆ ˆ ˆ ˆ the coordinates [a− . (e) Find a simple way (avoiding putting in the value of e) to calculate the Bohr magneton µe = e¯ /2me and the nuclear magneton µp = e¯ /2mp in electronvolt per Tesla (eV/T ). a2 ] need the basis vectors n− . 0) ˆ n2 = (0.

2) Acting on the wave function one finds for a free particle.4) where M is the particle mass. t) = exp(−i k 0 t + i k · r).1 and 2.3) Equations 2. ∂t 2M (2. 8 . t) = − ψ(r. i.e. ∂t2 (2. 2M (2.6) with (k 0 )2 = k2 + M 2 . we just want to play a bit with covariant equations and study their behavior under Lorentz transformations. t) = 0.1) under the substitution (in coordinate representation) E −→ Eop = i ∂ .Chapter 2 Relativistic wave equations 2. 2 + M 2 φ(r. Substitution of operators gives the Klein-Gordon (KG) equation for a real or complex function φ. E = p2 . i ∇2 ∂ ψ(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. (2. • The energy spectrum is not bounded from below: considering the above stationary plane wave solutions one obtains (2.1 The Klein-Gordon equation In this chapter. (2. written as pµ −→ i∂µ is covariant (the same in every frame of reference). The Schr¨dinger equation in quantum mechanics is the operator equation o corresponding to the non-relativistic expression for the energy. ∂t p −→ pop = −i∇. But the replacement 2. φk (r.5) Although it is straightforward to find the solutions of this equation. t). p 2 = p µ p µ = E 2 − p2 = M 2 . namely plane waves characterized by a wave number k. Thus a covariant equation can be obtained by starting with the (covariant) equation for the invariant length of the four vector (E. p). t) = ∂2 − ∇2 + M 2 φ(r.7) k 0 = ± k2 + M 2 = ±Ek . there are solutions with negative energy.3 are not covariant. (2.

j) of the four-current j. k) ≡ a(k) and φ(−Ek .Relativistic wave equations 9 • 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.13) ˜ with (in principle complex) coefficients φ(k). it is simply a matter of being careful to find a consistent (covariant) theory. ↔ ↔ (2. 2. (2π)3 2Ek (2. −k) ≡ b∗ (k) one has φ(x) = d3 k e−i k·x a(k) + ei k·x b∗ (k) = φ+ (x) + φ− (x). but rather the zero component of a four vector. The appropriate current corresponding to the KG equation (see Excercise 2. however. having the KG equation as a space-time symmetric (classical) equation. The KG equation. . The coefficients a(k) and b∗ (k) are the amplitudes of the two independent solutions (two. the position r simply becomes a number (parameter) on which the operator depend. 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). (2.5). ∂t (2.3).12) It is easy to see that this current is conserved if φ (and φ∗ ) satisfy the KG equation. relativistically the density is not a scalar quantity.11) Therefore. there are no longer fundamental objections to mix up space and time. As we will see both problems are related and have to do with the existence of particles and antiparticles. (2. for which we need the interpretation of φ as a field that must be quantized.14) ˜ ˜ Introducing φ(Ek . This leads to the existence of negative densities.8) (2.15) In Eqs 2. (2. 2M 2M ∂ρ = −∇ · j. φ(x) = d3 k (2π)3 2Ek ˜ ˜ e−i k·x φ(Ek . j) = i φ∗ ∂0 φ. 2. we note that an arbitrary solution for the field φ can always be written as a superposition of plane waves.15 one has elimated k 0 and in both equations k · x = Ek t − k · x. ∂µ j µ = 0. k) + ei k·x φ(−Ek . They are referred to as mode and anti-mode amplitudes (or because of their origin positive and negative energy modes). It is possible to rewrite it as an integration over positive energies only but this gives two terms (use the result of exercise 2. very much similar as the time t. φ(x) = d4 k ˜ 2π δ(k 2 − M 2 ) e−i k·x φ(k) (2π)4 (2.14 and 2. after restricting the energies to be positive). which is what Lorentz transformations do.2 Mode expansion of solutions of the KG equation Before quantizing fields.9) They satisfy the continuity equation. Then.10) which follows directly from the Schr¨dinger equation. Besides a wave function the field must be an operator. −k) . This continuity equation can be written o down covariantly using the components (ρ. is a second order equation and φ and ∂φ/∂t can be fixed arbitrarily at a given time. −i φ∗ ∇ φ . And. Its argument. The integration over k-modes clearly is covariant and restricted to the ‘mass’-shell (as required by Eq.2) is j µ = i φ∗ ∂ µ φ ↔ or (ρ.

Performing some Lorentz transformation x → x′ = Λx. 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) . It is trivial to see that ∂µ ∂ µ + M 2 φ∗ (x) = 0.Relativistic wave equations 10 2. Since a · b = a · ˜ it is easy to see that ˜ b. for which we consider space inversion.21) d3 k x x e−i k·˜ a(k) + ei k·˜ b∗ (k) (2π)3 2Ek d3 k ˜ ˜ e−i k·x a(k) + ei k·x b∗ (k) (2π)3 2Ek d3 k e−i k·x a(−k) + ei k·x b∗ (−k) . x Transforming everywhere in the KG equation x → x one obtains ˜ ˜ ˜ x ∂µ ∂ µ + M 2 φ(˜) = 0. This shows how charge conjugation transforms ’particle’ modes into ’antiparticle’ modes and vice versa. φP (x) ≡ φ(˜) (P for parity). This shows how parity transforms k-modes into −k modes.23) i.18) (2.19) (2. (2π)3 2Ek (2.e. (2.20) one has for the mode amplitudes aP (k) = a(−k) and bP (k) = b(−k). x) → (t.16) The consequence of this is discussed in Exercise 2. −x) ≡ (˜µ ).6 We will explicitly discuss the example of a discrete symmetry. for the mode amplitudes aC (k) = b(k) and bC (k) = a(k). 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) .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. (2π)3 2Ek (2. changing the sign of the spatial coordinates. (2π)3 2Ek (2. (2. x ∂µ ∂ µ + M 2 φ(˜) = 0. i. For the real field one has aC (k) = a(k).17) implying that for each solution φ(x) there exists a corresponding solution with the same energy.22) showing that with each solution there is a corresponding charge conjugated solution φC (x) = φ∗ (x). (2. Another symmetry is found by complex conjugating the KG equation. .e. which implies (xµ ) = (t. one thus must have that the function φ → φ′ such that φ′ (x′ ) = φ(x) or φ′ (x) = φ(Λ−1 x).

the following property of delta functions δ (f (x)) = X 1 δ(x − xn ). letting V → ∞. Q = 0.2) in terms of the a(k) and b(k) using the expansion in Eq.3 Show that1 d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) F (k 0 .1) for a complex scalar field current (exercise 2. ˙ and thus for any normalized solution the full ‘charge’. Use these to show that derivative i∂0 φ(k. 2. Exercise 2. t).1 Show that for a conserved current (∂µ j µ = 0) the charge in a finite volume. k). that j µ = i φ∗ ∂ µ φ is a conserved current Note: A ∂µ B ≡ A∂µ B − (∂µ A)B). S V d3 x j 0 (x).5 Express the full charge QV (exercise 2. k) = (2π)4 k2 + M 2 . (2π)3 2Ek where Ek = Exercise 2. t) ≡ d3 x φ(x. d3 k F (Ek . Similarly express the quantities E= Pi = 1 Use d3 x (∂0 φ)∗ (∂0 φ) + ∇φ∗ · ∇φ + M 2 φ∗ φ . QV ≡ satisfies ˙ QV = − ds · j.15.4 ˜ write down the 3-dimensional Fourier transform φ(k. t).15. t) exp(−i k · x) and its time ˜ t) for the mode expansion in Eq. ˜ b(k) = ei Ek t i ∂0 φ∗ (−k. ↔ ↔ Exercise 2.2 Show that if φ and φ∗ are solutions of the KG equation. ↔ ↔ Exercise 2. t) = ei Ek t (i∂0 + Ek ) φ(k. Note that a(k) and b(k) are independent of t. ˜ ˜ a(k) = ei Ek t i ∂0 φ(k. d3 x (∂ {0 φ)∗ (∂ i} φ).Relativistic wave equations 11 Exercises Exercise 2. 2. is conserved. |f ′ (xn )| zeros xn .

(a) Show that relativistic invariance.7 To solve the problem with positive and negative energies and get a positive definite density. and β 2 = I. t) = (−i α · ∇ + m β) ψ(r. (b) Show that one has current conservation with as current j 0 = ψ†ψ and j = ψ † αψ.Relativistic wave equations 12 in terms of the a(k) and b(k). 1 2 Exercise 2. where k ′ = Λ−1 k with k = (Ek . 2 We denote the commutator as [A. i ∂ ψ(r. but realize which ones are common to the calculations. Note: Do not repeat all steps. As an intermediate step you could use (check!) for two field φ1 (with coefficients a1 and b∗ ) and φ2 (with coefficients 1 a2 and b∗ ) the result 2 d3 x φ∗ (x) φ2 (x) 1 = d3 k 2 (2π)3 4Ek a∗ (k)a2 (k) + b1 (k)b∗ (k) 1 2 + a∗ (k)b∗ (−k) e2iEk t + b1 (k)a2 (−k) e−2iEk t . B} = AB + BA. .e. ∂t where ψ is a multi-component wave function and α and β are matrices in that wave function space. β} = 0. a{µ bν} ≡ aµ bν + aν bµ . k). {αi . We will encounter these quantities later as energy and momentum. Note that a{µ bν} indicates symmetrization. B] = AB − BA and the anticommutator as {A. Exercise 2. t). αj } = 2 δ ij I. making sure that each component of ψ satisfies the KleinGordon equation requires the anticommutation relation2 {αi . i. one can try to construct a first order differential equation for the time evolution.6 Write down the mode expansion for the Lorentz transformed scalar field φ′ (x) and show that it implies that the Lorentz transformed modes are a′ (k) = a(k′ ).

Before proceeding with the Lorentz group we will first discuss the rotation group as an example of a Lie group with which we are familiar in ordinary quantum mechanics. 3. in which the parameter space forms locally a Euclidean space. provided we identify the antipodes. since the symmetry group describing rotations is embedded in the Lorentz group. Any rotation can be uniquely written as R(θ.Chapter 3 Groups and their representations Simple systems that ought to be described with relativistic equations are free particles with spin.g. two-component fields in analogy to the two-component wave functions used to include spin in a quantum mechanical description of an electron in the atom. we must study the representations of the Lorentz group. z )V . The KG equation will actually remain valid. In this section we will investigate if there exist objects other than just a scalar (real or complex) field φ. it also must hold for particles with spin. n) where n is a unit vector and ˆ ˆ θ is the rotation angle. e. Ly =  0 0       0 i 0 −i 0 13 that are generated by  i    0 . However.and y-axes     0 0 0   0 0       Lx =  0 0 −i  . For an infinitesimal rotation around the z-axis one has ˆ R(δθ.     0 0 0 (3. electrons. n) ≡ R(π. Since the KG equation expresses just the relativistic relation between energy and momentum. z ) = 1 + i δθ Lz with as generator ˆ 1 ∂R(θ. 0 ≤ θ ≤ π. A general rotation — we will consider SO(3) as an example — is of the form   ′    sin θ 0   Vx   Vx   cos θ   ′    V  =  − sin θ cos θ 0   V   y   y    (3.3) θ=0 In the same way we can consider rotations around the x. The parameter-space of SO(3) is a ˆ ˆ ˆ sphere with radius π. in particular each component of these spinors will satisfy this equation.2) The rotation groups SO(3) and SU (2) are examples of Lie groups. e. −n).1 The rotation group and SU (2) for a rotation around the z-axis or shorthand V ′ = R(θ. Locally this parameter-space is 3-dimensional and correspondingly one has three generators. z ) Lz = i ∂θ    0 −i 0      =  i 0 0 .e.g. R(π.4) .1)        ′    0 0 1 Vz Vz (3. that is groups characterized by a finite number of real parameters.   0 (3. Particles with spin then will be described by certain spinors. i.

y] = −[y. or (Lk )ij = −i ǫijk . ˆ R(θ. Writing a0 = cos(θ/2) and a = n sin(θ/2) we have A = A(θ. The infinitesimal generators of SU (2) are obtained by ˆ considering infinitesimal transformations. It is straightforward to check that the conditions require    a0 + i a3 +i a1 + a2    = a0 1 + i a · σ A =  (3.e. z ) = lim R θ ˆ . n) = ˆ = 1 cos θ ˆ + i (σ · n) sin 2 θ ˆ exp i σ · n 2 θ 2 (3. thus is as the surface of a sphere in 4 Euclidean dimensions. ˆ therefore. for fixed n. the rotations in SO(3) can be generated from infinitesimal rotations that can be expressed in terms of a basis of three generators Lx . It is straightforward to check that any (finite) rotation can be obtained from a combination of infinitesimal rotations. y] ∈ A. y]] = 0 (Jacobi identity). namely that there exists a bilinear product [. ˆ ˆ A(θ. • [x. x]).Groups and their representations 14 (n.5) Rotations in general do not commute. [y. π ) π -sphere (antipodes identified) 2 π -sphere (surface identified) Figure 3. z]] + [y. These generators form a threedimensional Lie algebra SO(3).z N N N →∞ . we turn to the group SU (2) of special (det A = 1) unitary (A† = A−1 ) 2 × 2 matrices.1: the parameter spaces of SO(3) (left) and SU (2) (right). which reflects itself in the noncommutation of the generators. Locally this is a 3-dimensional Euclidean space and SU (2). but with all points at the surface identified (see figure).2 π) (-n. y ∈ A ⇒ [x. (3. (3. [x. x]] + [z. thus. With matrix commutation this algebra satisfies the requirements for a Lie algebra. is a 3-dimensional Lie-group. i. π) (n. They satisfy the commutation relations [Li . x] = 0 (thus [x. (3. These matrices can be defined as acting on 2-component spinors (χ → Aχ) or equivalently as acting on 2 × 2 matrices (B → ABA† ). for rotations around z for instance. also can be considered as a filled 3-sphere with radius 2π.6) Summarizing. One way of viewing the parameter space. ] that satisfies • ∀ x. [z.7)  +i a1 − a2 a0 − i a3          1 0   + i a1  0 1  + i a2  0 −i  + i a3  1 0        = a0  (3.10) . • [x.9) 3 The parameter-space. Lj ] = i ǫijk Lk .8)         0 1 1 0 i 0 0 −1 with real a’s and i=0 (ai )2 = 1. Next. n) ≈ 1 + i θ J · n. Ly and Lz .

J− = :0. the above mapping corresponds to the trivial mapping of the Lie algebras. . since any finite rotation can be constructed from the infinitesimal ones. SU (2) ≃ SO(3). These are mappings from G into a finite dimensional vector space (its dimension is the dimension of the representation). however. From the algebra one derives J 2 |j. . −j one has for j = 0: 8 9 8 9 8 9 Jz = :0.11) Thus σx /2. m = j(j + 1) − m(m ± 1)|j. Jz |j.14) For a Lie group. (3. n) −→ R(θ. In order to find local representations Φ of a Lie-group G. one has a Lie algebra isomorphism that is linear and preserves the bilinear product. (3. σy /2 and σz /2 form the basis of the Lie-algebra SU (2). j − 1. (3. it is sufficient to consider the representations Φ of the Lie-algebra G. n) i ∂θ = θ=0 σ ˆ · n. the SU (2) → SO(3) mapping is a 2 : 1 mapping where both A = ±1 are mapped into R = I. j − 1. n) ˆ −→ R(2π − θ. m . In the full parameter space. Casimir operators commute with all the generators and the eigenvalue of J 2 can be used to label the representation (j). −j. The existence of symmetries means that there are operators g belonging to a symmetry group G that commute with the Hamiltonian. [g. preserving the group structure. m = m|j. which preserve the Lie-algebra structure. i. sometimes in more than one way (but this will be discussed later). Explicit representations using the basis states |j. ] = i ǫijk . The other generators Jx and Jy (or J± ≡ Jx ± iJy ) then transform between the states in V (j) .13) The relation can be used to establish the homomorphism (Check that it satisfies the requirements of a homomorphism). . J+ = :0. .15) Representations Φ of a group are mappings of G into a finite dimensional vector space. i. . [g. Among the generators one looks for a maximal commuting set of operators (in this case consisting of the operator Jz and the (quadratic Casimir) operator J 2 ). . 2 2 2 (3. the commutation relations. m ± 1 with 2j + 1 being integer and m = j. H] = 0 for g ∈ G.Groups and their representations 15 with ˆ J ·n≡ ˆ 1 ∂A(θ. .e. H] = 0 for g ∈ G.12) One. . They satisfy [ σk σi σj .2 Representations of symmetry groups The presence of symmetries simplifies the description of a physical system. m = j(j + 1)|j. Suppose we have a system described by a Hamiltonian H. it is sufficient that the generators commute with H. n) A(σ · a)A−1 = σ · RA a 0≤θ≤π π ≤ θ ≤ 2π.e. m = j. . while J± |j. 2 (3.e. immediately sees that the Lie algebras are identical. m with m-values running from the heighest to the lowest. 3. m . There exists a corresponding mapping of the groups given by µ: SU (2) −→ SO(3) ˆ ˆ A(θ. . i. Near the identity. m with eigenvalues of Jz . thus. . Within the (2j + 1)-dimensional representation space V (j) one can label the eigenstates |j.

0 0 0 √ 2 0 0 0 0 9 > > > > >. U (φ. Jz = > i > > > > > > > > > > : . This turns out to be the case for all half-integer representations of the Lie algebra. > > . however. one obtains the unitary representations Φ(g) = exp(i Φ(g)) of G.16) (3. > > > > . 0 0 0 √ 2 for j = 1: If rotations leave H invariant. For SU(2) the conjugate representation is not a new one. Given a representation.3). . the possibility thus exist that some extensions will not be well-defined representations. 1 0 0 √ 2 0 √ 9 0 > >. J− = > 2 > > > 0 > > > > > : . i. Of all groups of which the Lie algebras are homeomorphic. all states in the representation space have the same energy. 0 0 √ 3 0 0 0 0 √ 2 0 0 0 0 9 > > > > > >. the conjugate transformation U ∗ acts on χ∗ . θ. 2 −i 0 0 0 0 0 9 > > > >. The jump U → U ∗ implies for the generators σ/2 → −σ ∗ /2. 0 −1/2 √ 2 0 0 8 9 1 > > > . one can look at the conjugate representation.17) By moving through the parameter space the D-functions can be extended to global functions for all allowed angles.g. 1 |1. θ. any point in the parameter space can be reached in a unique way and any local (infinitesimal) representation can be extended to a global one. (j) (j) ′ The spin matrices for that Cartesian basis {ǫx . : 0 i 0 −i 0 0 0 (j) (3. however. m = Dm′ m (φ. J+ = > 0 > > : . e. χ) = eim φ dm′ m (θ) e−imχ . then ǫχ∗ → −ǫ σ∗ χ∗ = σǫχ∗ ]. For an element in the group G the corresponding point in the parameter space can be reached in two ways. . 2 |1. Therefore the representation and conjugate representation are equivalent in this case (see Excercise 3. The matrix elements of the unitary representations are known as D-functions.Groups and their representations 16 for j = 1/2: 8 > 1/2 Jz = > : 0 0 0 0 0 0 −1 8 9 0 > > > . however. > . J− = > 2 > > : . For those global representations. for an element A of SU (2) or R of SO(3) parametrized with Euler angles. ǫy . : 0 2 0 0 0 0 √ 3 0 0 8 9 > √ > > 0 > > > > > > > > . dm′ m (θ) ≈ δm′ m − iθ(J 2 )m′ m .e. Consider the j = 1/2 representation of SU(2). SU (2) is the covering group of SO(3). Jy = > 0 0 0 > . For SO(3). The group SO(3). 1 ≡ ǫ1 ≡ − √ (ǫx + i ǫy ) . 0 ≡ ǫ0 ≡ ǫz . 0 8 9 > √ > > 0 > > > > > > . is not simply connected. to be precise the states ǫχ∗ transform via σ. 0 From the (hermitean) representation Φ(g) of G. If the group is simply connected. Because there exists a matrix ǫ such that σ = −ǫ σ∗ ǫ−1 one immediately sees that appropriate linear combinations of conjugate states. −1 ≡ ǫ−1 ≡ √ (ǫx − i ǫy ) . 0 0 −1/2 0 0 0 0 −3/2 8 9 > 0 > > > > > > >. [Explicitly. θ. however. the simply connected group is called the (universal) covering group. For a decent (welldefined) global representation. if χ → σχ and χ∗ → −σ ∗ χ∗ . This works for SU (2). the extension from a local one must be unique. J+ = > 0 : . Infinitesimally (around the identity) the D-functions for SU (2) and SO(3) are the same. ǫz } are 8 9 8 9 8 > 0 0 0 > > 0 0 i > > 0 > > > > > > > > > > Jx = > 0 0 −i > . For j = 1 another commonly used representation starts with three Cartesian basis states ǫ related to the previous basis via 1 |1. J− = > 0 : . J = > 0 > > + > > > > 0 > > > > . or equivalently the Hilbert space can be written as a direct product space of spaces V (j) . j. and for j = 3/2: 8 0 > 3/2 > > 0 > 1/2 > Jz = > > 0 > 0 > : 0 0 8 > 1 > > > Jz = > 0 > : 0 8 9 > 0 > > > > > > > > . χ) = e−iφJz e−iθJy e−iχJz . that is any closed curve in the parameter space can be contracted to a point. χ)|j. the topological structure of the group is important. : . There exist two different types of paths. contractable and paths that run from a point at the surface to its antipode. m′ |U (φ. If a transformation U acts on χ.

(3.     0 0   g00 g01 g02 g03   1 0      g   0 −1 0   g11 g12 g13   0  =      G =  10 (3. Thus one has the (pseudo-orthogonality) relation ΛT GΛ = G ⇔ GΛT G = Λ−1 ⇔ ΛGΛT = G. 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 L↑ + L↓ + L↑ − L↓ − proper orthochrone proper non-orthochrone improper orthochrone improper non-orthochrone .22) = Λµν and (G) µν = gµν gives (3.24) (ii) |Λ00 | ≥ 1. e.  20 g21 g22 g23            g30 g31 g32 g33 0 0 0 −1 the scalar product can be written as T x2 = xT G x (3. it is easy to derive some properties of the matrices Λ: (i) det(Λ) = ±1.23) (ΛT )ρµ (G)µν (Λ)νσ = (G)ρσ . det Λ = −1 is called improper). it is convenient to use a vector notation for the points in Minkowski space. Denoted in terms of column vectors and 4-dimensional matrices one writes for the poincar´ transe formations x′ = Λ x + a.19) (Note that x is a row vector).20) The proper tensor structure of the matrix element (Λ)µν is a tensor with one upper and one lower index. σ/2 in a two-dimensional (spin 1/2) case. e To derive some of the properties of the Lorentz group. We considered the generators and looked for representations in finite dimensional spaces.21) (3.3 The Lorentz group In the previous section spin has been introduced as a representation of the rotation group SU (2) without worrying much about the rest of the symmetries of the world. (det Λ = +1 is called proper. 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 . proof: (ΛT GΛ)00 = (G)00 = 1 → Λµ0 gµν Λν0 = 1 → (Λ00 )2 − i (Λi 0 )2 = 1. explicitly x′µ = (Λ)µν xν + aµ or x′µ = Λµν xν + aµ .18)  g   0 0 −1 0  . (3. From this property. consisting of the Lorentz group and translations.g.Groups and their representations 17 3. Writing x as a column vector and the metric tensor in matrix form. proof: det(ΛT GΛ) = det(G) → (det Λ)2 = 1. which as a matrix equation with (Λ) µν (3. In this section we consider the Poincar´ group.

29)  0 −   0 −1 0      0 0 0 −1 . J 3 ). Rotations around the z-axis are given by ΛR (θ. (3. This is the origin of the Thomas precession. Thus ˆ  0′       0 0 0  V 0   V   1  0 0 0 0         V 1′   0 cos θ sin θ 0   V 1   0 0 −i 0                      −→ J3 =   V 2′  =  0 − sin θ cos θ 0   V 2   0 i 0 0  . Returning to the global group. 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).27)  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 . Using these explicit transformations. [J i . K j ] = i ǫijk K k . In L↑ . since I ∈ L↑ and + + det Λ and Λ0 change continuously along a path from the identity. The first two sets of commutation relations exhibit the rotational behavior of J and K as vectors under rotations.Groups and their representations = i=1 (Λ0 )2 . infinitesimally given by ˆ ΛR (θ.25)                 3′    3    V 0 0 0 1 V 0 0 0 0 Boosts along the z-direction are given by ΛB (φ. z ) = exp(i θ J 3 ). J 2 . 3 i 2 i=1 (Λ0 ) 3 18 (iii) Note that Lorentz transformations generated from the identity must belong to L↑ .   0 0   +1 0    0 +1 0   0   ∈ L↑ (identity)   I =  (3. and those of the boosts. which satisfy the commutation relations (check!) [J i . since the generators K do not form a closed algebra. [K i . K 2 . K = (K 1 . From the commutation relations one sees that the boosts (pure Lorentz transformations) do not form a group. we have found the generators of rotations. i proof: use ΛT GΛ = G and ΛGΛT = G. J = (J 1 . βγ = sinh φ. Thus  0′    0     V   cosh φ 0 0 sinh φ   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  .26)         0 0 1 0         3′    3    V sinh φ 0 0 cosh φ V i 0 0 0 The parameter φ runs from −∞ < φ < ∞. The commutator of two boosts in different directions (e. z ) ≈ ˆ ˆ I − i φ K 3 .g. 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 φ as γ = cosh φ. (3. one distinguishes rotations 0 + and boosts. it is easy to find the following typical examples from each of the four classes. K 3 ). infinitesimally given by ΛB (φ. K j ] = −i ǫijk J k . z ) ≈ I + i θ J 3 . z ) = exp(−i φK 3 ).28)  0 −   0 +1 0      0 0 0 +1   0 0   +1 0    0 −1 0   0   ∈ L↑ (space inversion)   Is =  (3.

In order to find the commutation relations including the translation generators. 3.32) thus (ω)ij = −(ω)ji and (ω)0i = (ω)i0 . K 3 = M 03 ). J 3 = M 12 ) and for the (infinitesimal) boosts (along (3. M ρσ ] = −i (g µρ M νσ − g νρ M µσ ) − i (g µσ M ρν − g νσ M ρµ ) .30) These four transformations form the Vierer group (group of Klein). Multiplying the proper orthochrone transformations with one of them gives all Lorentz transformations.g. which are obtained by writing the infinitesimal parameters (ω)µν in terms of six antisymmetric matrices (M αβ )µν . The e 1 These ΛB = I − i φ3 K 3 = I − i ω30 M 30 .34) The algebra of the generators of the Lorentz transformations can be obtained by an explicit calculation1 . we have for the (infinitesimal) rotations (around z-axis) ΛR = I + i θ3 J 3 = I − i ω12 M 12 .4 The generators of the Poincar´ group e ↑ The transformations belonging to P+ are denoted as (Λ. The extension to the Poincar´ group is e ↑ straightforward.g. the Lorentz transformations form a (Lie) group (Λ1 Λ2 and Λ−1 are again Lorentz transformations.36) (e. This is a + normal subgroup and the factor group is the Vierer group. ǫ). One can easily convince oneself that i (ω)µν = ω µν = − ωαβ (M αβ )µν . commutation relations are the same as those for the ’quantummechanics’ operators i(xµ ∂ ν − xν ∂ µ ) . [M µν . Also this group can be divided into four parts. There are six generators.Groups and their representations  0 0  −1 0   0 −1 0  0  =   0  0 −1 0   0 0 0 −1       ∈ L↓   +    19 Is It (space-time inversion) (3. explicitly reading x′µ = (Λ)µν xν + aµ = (δ µν + (ω)µν ) xν + ǫµ = Λµν xν +a = µ inf (3. 1 with θi = − 2 ǫijk ω jk and J i = z-axis). P+ etc. Thus the two vector operators J and K form under Lorentz transformations an antisymmetric tensor M µν .31) (g µν + ω µν ) xν +ǫ . (3. We now want to find a covariant form of the six generators of the Lorentz transformations. We will just require Poincar´ invariance for the generators themselves. µ The condition ΛT GΛ = G yields g ρµ + ω ρµ gρσ (g σν + ω σν ) = gµν =⇒ ωνµ + ωµν = 0. Of the four parts only L↑ forms a group.37) with φi = ω i0 and K i = M 0i (e.33) (3.35) Explicitly. we continue using covariance arguments. 2 (M αβ )µν = i g αµ g βν − g α g βµ ν (3. Summarizing. (3. a). We therefore find (again) that there are six generators for the Lorentz group. infinitesimally approximated by (I + ω. 1 ijk M jk 2ǫ (3. three of which only involve spatial coordinates (rotations) and three others involving time components (boosts).

that commute e among themselves. For instance. [J i . J j ] = i ǫijk J k . because one then sees that by letting c → ∞ the commutation relations of the Galilei group. [K i . the generators J 2 and J 3 in the case of the rotation group. P j ] = i ǫijk P k . check that U1 U2 corresponds with Λ1 Λ2 ).42) Note that the commutation relations among the generators M µν in Eq. = pµ |p. 3.41) (3. known from non-relativistic quantum mechanics are obtained. = −i (g µρ P ν − g νρ P µ ) . P j ] = [P i . P ρ ] = 0. (3. (3. a)P µ U (Λ. [P µ . a)P µ U † (Λ. The generators of the translations are the (momentum) operators P µ . (3.45) . For instance.39) (to decide on where the inverse is. [J i . To find other generators that commute with P µ we look for Lorentz transformations that leave the four vector pµ invariant.44) = = = pµ + ω µν pν = pµ 0 ǫµνρσ pρ sσ . . Infinitesimally. writing the generator P = (H/c. 3. (3. They just state that M µν transforms as a tensor with two Lorentz indices.43) [K i . . a) = (Λ−1 )µν P ν . These form a group called the little group associated with that four vector. P j ] = i δ ij H/c2 . [K i .35 could have been obtained in the same way. one obtains 2 [P i . a) = (Λ)µν P ν or U † (Λ. The eigenvalues of these will be the four-momentum pµ of the state. The requirement that P µ transforms as a four-vector (for which we know the explicit behavior from the defining four-dimensional representation). P µ |p. ǫ) = 1 − ωαβ M αβ + i ǫα P α . (3. as part of the set. . P ) in terms of the Hamiltonian and the three-momentum operators. the generators J± in the case of the rotation group. K j ] = i ǫijk K k . the generators M αβ no longer exclusively act in Minkowski space. Λµν pν ⇒ ωµν p ⇒ ωµν ν In order to label the states in an irreducible representation we construct a maximal set of commuting operators. [J i . . leads to U (Λ. K j ] = −i ǫijk J k /c2 . These define states with specified quantum numbers that are eigenvalues of these generators. Explicitly. H] = [J i . P ν ] [M µν . .Groups and their representations 20 infinitesimal form of any representation of the Poincar´ group (thus including the translations) is of e the form i U (I + ω. H] = i P i .5 Representations of the Poincar´ group e where pµ is a set of four arbitrary real numbers. (1 + iǫγ P γ − i i ωαβ M αβ )P µ (1 − iǫδ P δ + ωρσ M ρσ ) = P µ + ω µν P ν . H] = 0. Taking any one of the states in an irreducible representation.38) 2 At this point. . 2 2 (3. Of the generators of the Poincar´ group we choose first of all the generators P µ . other states in that representation are obtained by the action of operators outside the maximal commuting set. the tensor M µν in terms of boosts cK i = M 0i and rotations J i = 1 ǫijk M jk . We have here reinstated c. . (3.40) giving the following commutation relations by equating the coefficients of ǫµ and ωµν .

Groups and their representations 21 where s is an arbitrary vector of which the length and the component along pσ are irrelevant. 2 (3. . for which we distinguish the following cases p 2 = m2 > 0 p2 = 0 pµ ≡ 0 p 2 = m2 > 0 p2 = 0 p2 < 0 p0 > 0 p0 > 0 p0 < 0 p0 < 0 Only the first two cases correspond to physical states. These e operators are the Casimir operators of the algebra and can be used to define the representations of ↑ P+ .47) where the Pauli-Lubanski operators W µ are given by 1 Wµ = − ǫµνρσ M νρ P σ . . . . = = i exp − ǫµνρσ M µν pρ sσ |p. Given the momentum four vector pµ we choose a tetrade consisting of three orthogonal spacelike unit vectors ni (p).53) . . The elements of the little group thus are U (Λ(p)) = = and U (Λ(p))|p. . 0.49) (3. M ǫijk . Wα ] = [Wµ .51) This will enable us to pick a suitable commuting ’spin’ operator. From the algebra of the generators one finds that P 2 = Pµ P µ and W 2 = Wµ W µ (3.48) The following properties follow from the fact that W µ is a four-vector by construction. .50) (3. (3. Massive particles: p2 = M 2 > 0. Pν ] = [Wµ . the third case represents the vacuum. Wν ] = −i (gµα Wν − gνα Wµ ) .52) commute with all generators and therefore are invariants under Poincar´ transformations. i ǫµνρσ W ρ P σ . 0.46) (3. satisfying gµν nµ (p)nν (p) = j i nµ (p) pµ i ǫµνρσ pµ nν nρ nσ i j k = = −δij . i exp − ωµν M µν 2 i exp − ǫµνρσ M µν pρ sσ 2 (3. . of which the components generate the little group of pµ or (for the third of the following relations) from explicit calculation [Mµν . 2 exp (−isµ W µ ) |p. . (3. while the others have no physical significance. p0 > 0. which can thus be chosen spacelike.

. (3. (1. Together with the four momentum states thus can be labeled as |M.e. 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. In that case the vectors ni (p) are just the space directions and W = (0. it is sufficient to choose a particular frame to investigate the coefficients in Eq. S j (p)] = −i nµ (p)nν (p) j i ǫµνρσ W ρ P σ = i ǫijk S k (p). n1 (p). 0. show that W /M can be interpreted as the intrinsic spin. s.54) (i.e. n2 (p0 ) ≡ s(p0 ) ≡ (0. the S i (p) form the generators of an SU (2) subgroup that belongs to the maximal set of commuting operators. 1.57) M2 i i and using the completeness relation X i « „ pµ pν . 1 2 (3. |p3 |).62) . 0. p. Having made a covariant decomposition.54 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..58) one sees that S i (p) ”2 =− W2 . ms . . 0. 1. 0. −1) (3. nµ (p)nν (p) = − g µν − i i M2 X“ i (3.56) (3. . M J) with components W i = (M/2)ǫijk M jk .60) where E = p2 + M 2 and ms is the z-component of the spin −s ≤ ms ≤ +s (in steps of one). p0 > 0. 0). 3. The commutation relations [W i (p). 0). In this case a set of four independent vectors is chosen starting with a reference frame in which pµ = pµ 0 in the following way: p0 = (p0 . 0. More generally (fully covariantly) one can proceed by defining S i (p) ≡ W i (p)/M and obtain [S i (p). The explicit construction of the wave function can be done using the D-functions (analogous as the rotation functions).Groups and their representations 22 We can write W µ (p) = 3 W i (p) nµ (p) i i=1 (3. M2 (3. n1 (p0 ) ≡ (0.59) 1 Thus W 2 has the eigenvalues −M 2 s(s + 1) with s = 0. W j (p)] = i M ǫijk W k (p). 0.55) i. M2 (3. We will not do this but construct them as solutions of a wave equation to be discussed explicitly in chapter 4 Massless particles: p2 = M 2 = 0. 0). (3. W i (p) = −W · ni (p)). 1. 2 .61) such that in an arbitrary frame where the four vectors p. Noting that X X “ i ”2 1 W µ (p)W ν (p) S (p) = niµ (p)niν (p).

thus. 3. |p. is characterized by its momentum and the helicity. 1 2 23 (3. Thus we have P 2 . and gives [W 1 (p). −i W 1 (p). a)|0 = |0 .Groups and their representations The above does actually include the massive case. The meaning of λ(p) is most easily seen by comparing for a momentum eigenstate W 0 (p) = λ(p) |p| with Thus λ(p) = 1 W 0 (p) = − ǫ0ijk Mij pk = J · p. Show that if det(U ) = 1.70) (3. The vector W µ (p) now can be expanded W µ (p) = W 1 (p) nµ (p) + W 2 (p) nµ (p) + λ(p) pµ + W s (p) sµ (p). Exercises Exercise 3.66) We reproduce the algebra for the massive case with W 3 (p) = M λ(p).63) where W · p = 0 implies that W s (p) = −(M 2 /p · s) λ(p). λ . This state is in physical applications nondegenerate and is denoted by |0 . . The eigenvalues of W 1 and W 2 (in that case corresponding to the translations) thus can take any continuous values. W 1 (p)] [λ(p). i W 2 (p). which can take any integer or half-integer value.1 Show that for a unitary operator U = exp[iαk Jk ] with real coefficients αk . A massless particle. W 2 and λ is derived from the general commutation relations for [Wµ . note that the definition of eA is eA ≡ n=0 An /n!).67) as a commuting set with zero eigenvalues for P 2 and W 2 . W 2 (p)] = = = i M 2 λ(p). 2 J ·p . This algebra is (for instance) isomorphic to the one generated by rotations and translations in a 2-dimensional Euclidean plane (see exercises).64 has a vanishing right hand side). W µ (p) = λ(p) P µ . The algebra of W 1 .65) (3. the trace of the operator Jk is zero (Convince yourself that ∞ det(eA ) = eTrA by diagonalizing A. We don’t know of any physical relevance. P µ . and the eigenvalues of W 1 and W 2 are set to zero (corresponding to the limit M → 0 for the W µ Wµ eigenvalues of M 2 s(s + 1) → 0).69) One has P µ |0 = W µ |0 = 0 and the state is invariant under Lorentz transformations. But if M = 0 one has a different algebra (Eq. (3. implying continuous values for the spin (actually for W i = M S i ). the operators Jk must be hermitean. U (Λ. W 2 (3. Wν ]. however. |p| (3. W 2 (p)] [λ(p). The vacuum: pµ = 0.64) (3.68) which is called the helicity. (3. Note that angular momentum does not contribute to helicity as L · p = 0.

Groups and their representations 24 Exercise 3.3 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. Comment: this establishes that for spin 1/2 one doesn’t need to talk about ’anti-spin’ for the antiparticle. the action of RA b · RA a = = for any vectors a and b. From either of these.e. 3.64 to 3. consistent with the (canonical) commutation relations of a quantum theory. Excercise 3. Exercise 3.g.66 of the Pauli-Lubanski operators for massless particles. 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 Excercise 3.5 (optional) One might wonder if it is actually possible to write down a set of operators that generate the Poincar´ e transformations.13 A(σ · a)A−1 = σ · RA a ˆ (valid for any vector a). there is a one to one mapping) the algebra in Eqs 3. Note: Recall (or derive) that the Pauli matrices σ satisfy σi σj = δij + i ǫijk σk and Tr(σi σj ) = 2 δij . This is true for SU (2) but e. i.e.4 Consider rotations and translations in a plane (two-dimensional Euclidean space). it is not true for SU (3) (flavor or color properties of particles). derive the commutation relations and show that the algebra is isomorphic to (i. Tr(σi σj σk σl ) = 2 (δij δkl − δik δjl + δil δjk ) . Use a 3-dimensional matrix representation or consider the generators of rotations and translations in the space of functions on the (x-y) plane. show that one matrix ǫ exist such that ǫ† J ǫ = J c . Show that the Lie-algebra representations J and J c are equivalent. what are the matrices J c in the Lie-algebra representation describing the spin rotations for two-component ’anti-spinors’. (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). Tr(σi σj σk ) = 2i ǫijk . For a spin 1/2 antiparticle the representation A∗ is needed. Both particle and antiparticle have ’spin 1/2’.2 (a) Derive from the relation that defines the mapping of SU (2) matrices A into SO(3) matrices RA in Eq.. This is . where A = exp(i φ · σ/2) = exp(iφ σ · n/2). Writing A∗ = exp(i α · J c ).

Comment: extending this to more particles is a highly non-trivial procedure. Hint: for the Hamiltonian. you might just check relations involving J or K by taking some (relevant) explicit components. show first the operator identity [r. the others vanish. momentum and spin operators e satisfy the canonical commutation relations. f (p)] = i ∇p f (p). H P J K = p 2 c2 + m 2 c4 . pj ] = [ri . [ri . = p. [ri . Do this by showing that the set of operators. sj ] = 0. p×s 1 (rH + Hr) − t p + . These do not matter in the non-relativistic limit (c → ∞). although the presence of an interaction term V (r1 .Groups and their representations 25 possible for a single particle. but it can be done. pj ] = iδ ij and [si . that’s why many-particle non-relativistic quantum mechanics is ’easy’. r2 ) inevitably leads to interaction terms in the boost operators. . rj ] = [pi . = r × p + s. sj ] = [pi . if you don’t want to do this in general. sj ] = iǫijk sk . = 2c2 H + mc2 satisfy the commutation relations of the Poincar´ group if the position.

This tells us how to find the representations of the group. C) Instead of the generators J and K of the homogeneous Lorentz transformations we can use the (non-hermitean) combinations A = B which satisfy the commutation relations [Ai .8) ˆ ˆ ˆ with φ = φn and θ = θn.2) This shows that the Lie algebra of the Lorentz group is identical to that of SU (2) ⊗ SU (2). (j. where we restrict (for fixed n the parameters 0 ≤ θ ≤ 2π and 0 ≤ φ < ∞). B j ] = i ǫijk Ak . 2 1 (J − i K).6) (4.Chapter 4 The Dirac equation 4. 0) Type II : (0. K = i J (A = 0). C): Type I (denoted M ): Type II (denoted M ): 26 J= σ σ . j) K = −i J (B = 0). Aj ] = [B i . U (θ) H(φ) (4. It is simply connected and forms the covering group of L↑ . 2 (4.1 The Lorentz group and SL(2. 2 2 σ σ J = . C) is the group of complex 2 × 2 matrices with determinant one. similarly as the homeomorphism between SO(3) and SU (2). j ′ ).7) From the considerations above. (4. (4. M = exp i θ·σ 2 1 exp ± φ · σ 2 = U (θ) H(φ) . it also follows directly that the Lorentz group is homeomorphic with the group SL(2. 0. K = +i .3) (4. C). With this choice of parameter-spaces the plus and minus signs are actually relevant. K = −i . It is easy to see that these matrices can be written as a product of a + unitary matrix U and a hermitean matrix H.9) (4. respectively. They precisely correspond to the two types of representations that we have seen before.1) (4.4) (4. B j ] = [Ai . They will be labeled by two angular momenta corresponding to the A and B groups.5) = 1 (J + i K). becoming the defining representations of SL(2. Special cases are the following representations: Type I : (j. 2 2 (4. The group SL(2. i ǫijk B k .10) .

One defines spinors ξ and η transforming similarly under unitary rotations (U † = U −1 . (ǫ∗ = ǫ. U σ aµ U ˜ µ −1 H σ µ aµ H −1 = σ µ ΛB aµ H σ aµ H ˜ µ −1 (4. C). The Lorentz transformations Λ(M ) and Λ(M ∗ ). This is not true for the two-component spinors in SL(2. η → U η. (4. U ≡ (U † )−1 = U ) U (θ) = exp(i θ · σ/2) ξ → U ξ. With E = M γ = M cosh(φ) and p = M βγ n = M sinh(φ) n it is given by H(p) = exp φ·σ 2 = cosh φ 2 ˆ + σ · n sinh φ 2 = M +E+σ·p . H ≡ (H † )−1 = H −1 ). ˜ The sets of four operators defined by definitions σ µ ≡ (1. thus.18) As already discussed for SU (2). such that M = SM S −1 . (4. The following relations for + rotations and boosts are useful. are not covariant!) ˜ can be used to establish the homeomorphism between L↑ and SL(2. (4. −σ). C) and.15) (4. U σ µ aµ U −1 = σ µ ΛR aµ . η → Hη.14) satisfying Tr(σ µ σ ν ) = −2 gµν and Tr(σ µ σ ν ) = 2gµν = 2 δ µν (the matrices.e. ˜ µ Non-equivalence of M and M One might wonder if type I and type II representations are not equivalent. (4. Λ(ǫ) ∈ L↑ . Eqs 4. One has + −1 Λ(M ∗ ) = I2 Λ(M ) I2 . y They are connected via Λ(M ∗ ) = Λ−1 (ǫ)Λ(M )Λ(ǫ) within L↑ because iσ 2 = U (πˆ ) showing that + ǫ ∈ SL(2. σ µ ≡ (1.17) 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. Note that M ∗ = ǫ† M ǫ has its equivalent for the Lorentz transformations Λ(M ) and Λ(M ∗ ). ǫ−1 = ǫ† = ǫT = −ǫ). σ). ¯ ¯ H ∗ = ǫ† Hǫ = ǫ−1 Hǫ. C) or L↑ . This shows that M ∗ ≃ M . H(φ) = exp(−φ · σ/2).18 show that ±ǫξ ∗ transforms as a type-II (η) spinor. i.16) = σ Λ R aµ . if there does not exist a unitary matrix S.13) 2M (E + M ) Also useful is the relation H 2 (p) = σ µ pµ /M = (E + σ · p)/M . In fact. (4. C). namely H(φ) = exp(φ · σ/2). (4. thus.19) .12) A practical boost for the spinors is the one transforming from the rest frame to the frame with ˆ ˆ momentum p. 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). however. while ±ǫη ∗ transforms as a type-I + spinor. ˜ µ = σ Λ B aµ . cannot + be connected by a Lorentz transformation belonging to L↑ . ˜ (4. ξ −→ Hξ.11) but differently under hermitean boosts (H † = H. there exists the matrix ǫ = iσ 2 .The Dirac equation 27 Let us investigate the defining (two-dimensional) representations of SL(2.

therefore. they can be connected. a) and ˜ a = (a0 . + −   1 0   0 1   I2 =   0 0    0 0 0 0 −1 0 0 0 0 1      . they cannot be connected by a unitary 2 2 transformation. must combine the repe resentations.The Dirac equation 28 where The matrix I2 does not belong to L↑ . generators.17.   −→  M (Λ)  (4. Within the Lorentz group. m) =            χm uL H(p) 0 . but to L↑ . −→ −λ(p) −→ −K −→ H −→ J −→ t (scalar). i.  =      (4. where a = (a0 .13. etc. and the angular momentum in the rest frame is what we are familiar with as the spin of a particle. the boost in Eq. C) are suitable for representing spin 1/2 particles. 2 2 (4.      0   χm   uR   H(p)      . (scalar). The 2 2 angular momentum operators J i are represented by σ i /2. We also have seen that the representations (0. the aij elements of a tensor will not change sign. The representing member of the class P− is the parity µ µ or space inversion operator Is . m) = χm . Although these operators cannot be used to label the representations. 0) of SL(2. 0). The behavior is the same for classical quantities.20) 4.23) u(p. (axial vector). that have the correct commutation relations.1. A parity transformation changes a into a .2 Spin 1/2 representations of the Lorentz group Both representations (0. m) = η(0. Thus M ∗ and M are not equivalent within SL(2. Examples (with between brackets the Euclidean assigments) are r t p H J λ(p) K −→ −r (vector). ǫµνρσ will change sign.g. but by a transformation belonging ↑ ↑ to the class P− . 0) (via operators J and K) one sees that under parity 2 2 1 1 (0. −a). (pseudoscalar). This provides the easiest way of seeing ˜ what is happening. we obtain for the two components of u which we will refer to as chiral right and chiral left components. (vector). the well-known two-component spinor for a spin 1/2 particle.22) u=      η η M (Λ) 0 where M (Λ) = ǫM ∗ (Λ)ǫ−1 with ǫ given in Eq.e. but rather the operators W i (p) should be used. as we have seen in section 4. e. From the definition of the representations (0. For a particle at rest only angular momentum is important and we can choose ξ(0. 0) are inequivalent. 1 ) and ( 1 . 1 ) and ( 1 . i. consider the four component spinor that transforms under a Lorentz transformation as      0  ξ   ξ  . 1 ) and ( 1 . 4. we have seen that in the rest frame W i (p)/M → J i .e.      (4. For the spin 1/2 representations of the Poincar´ group including parity we. −→ −p (vector). 4. It has the same effect as lowering the indices. ) −→ ( .21) In nature parity turns (often) out to be a good quantum number for elementary particle states. C). Taking M (Λ) = H(p).

which in general is a linear equation in pµ . this can be rewritten 1 µ ν {γ . The explicit realization appearing in Eq.30) γ µ γ ν ∂µ ∂ν + M 2 ψ(x) = 0.27 is known as the Weyl representation.      0 H 2 (p)   uR   uR          = . .34) . As in section 2 we can use pµ = i∂µ to get the Dirac equation for ψ(x) = u(p)e−i p·x in coordinate space. Since ∂µ ∂ν is symmetric. = H −1 (p).7. 2 (4. (4. Applying (iγ µ ∂µ + M ) from the left gives (4.33) Multiplying with γ 0 on the right and pulling it through one finds ψ(x) iγ µ ∂µ +M ← = 0. γ } ∂µ ∂ν + M 2 ψ(x) = 0. It is of a form that we also played with in Exercise 2. 4.      (4. (pµ γ µ − M ) u(p) ≡ (/ − M ) u(p) = 0.27) which is a covariant (linear) first order differential equation. We first want to investigate if the Dirac equation solves the problem with the negative densities and negative energies. one must require that for ψ(x) the Klein-Gordon relation 2 + M 2 ψ(x) = 0 is valid for each component separately). (4. The general requirements for the γ matrices are thus easily obtained. (iγ µ ∂µ − M ) ψ(x) = 0. We will discuss another explicit realization of this algebra in the next section.    (4. The first indeed is solved. p (4. ψ † −iγ 0 ∂0 +iγ i ∂i −M ← ← = 0.The Dirac equation 29 with H(p) = ¯ H(p) = 2M (E + M ) E+M −σ·p 2M (E + M ) E+M +σ·p . γ ν } = 2 g µν .24) (4.26)       −2     H (p) 0 uL uL or explicitly in the socalled Weyl representation  −M E+σ·p       E−σ·p −M    uR      uL     = 0.28) To achieve also that the energy-momentum relation p2 = M 2 is satisfied for u(p).32) suppressing on the RHS the identity matrix in Dirac space. From this one obtains the Clifford algebra for the Dirac matrices. To see this consider the equation for the hermitean † † conjugate spinor ψ † (noting that γ0 = γ0 and γi = −γi ).25) It is straightforward to eliminate χm and obtain the following constraint on the components of u.29) which is an explicit realization of the (momentum space) Dirac equation. (4. (4.31) where γ µ are 4 × 4 matrices called the Dirac matrices. (4. {γ µ .

The Dirac sea forms the vacuum.38) The general algebra for the Dirac matrices is The Weyl (or chiral) representation:    0 1  0 1  .3 General representations of γ matrices and Dirac spinors {γ µ . opposite charge). γ ν } = 2 g µν .1: The Dirac sea of negative energy states and antiparticles. From the equations for ψ and ψ one immediately sees that j µ = ψγ µ ψ is a conserved probability current (exercise!). This problem was solved by Dirac through the introduction of a negative energy sea. Relying on the Pauli exclusion principle for spin 1/2 particles. but with properties such that it can be annihilated by the particle (e.The Dirac equation 30 E empty levels M -M filled levels E particle M -M hole Figure 4. γ =ρ ⊗1=  1 0 1 We Two often used explicit representations are the following1 : The standard representation:     σk   0  1 0  .  0 (4. 4.g.36) (4. (4. for the adjoint spinor ψ ≡ ψ † γ0 . Dirac supposed that the negative energy states were already completely filled and the exclusion principle prevents any more particles to enter the sea of negative energy states. γ k = iρ2 ⊗ σ k =  γ 0 = ρ3 ⊗ 1 =      0 −1 −σ k 0   0 γ = −iρ ⊗ σ =  k  σ k 2 k (4. remains.37) Using the explicit form of γ 0 in the Weyl representation (see Eq.35) is indeed positive and j 0 can serve as a probability density.40) .  −σ k  . one sees that there are two positive and two negative eigenvalues. We will see how this is properly implemented (also for bosons!) when quantizing fields.27).39) use ρi ⊗ σj with both ρ and σ being the standard 2 × 2 Pauli matrices. where γ 0 p0 ψ = M ψ −→ Eψ = M γ 0 ψ. The second problem encountered before. . E = +M (twice) and E = −M (twice). This hole forms an antiparticle with the same mass. From the vacuum a particle can be removed. This can most easily be seen in the particle rest frame. (4. The density j 0 = ψγ 0 ψ = ψ † ψ (4. the one of the negative energy states. 4.

10 are   1  σ3 0  3 12   = i [γ 1 . 4! (4.42) We note that the explicit matrix taking us from the Weyl representation to the standard representation.41) (4. 1 −1 2 For all representations one has † γµ = γ0 γµ γ0 .45) (γ5 )W. it is easy to see that PR/L = 1 (1 ± γ5 ) 2   1 0 =ρ ⊗1=  0 −1 3   . J =M =   0 σ3 2 4   1  −iσ 3 0   = i [γ 0 . γ σ ].R. (4. = ρ ⊗ 1 =   1 0 i ǫµνρσ µ ν ρ σ γ γ γ γ . K 3 = M 03 =    0 iσ 3 2 4 or in general one has the transformation i ψ → Lψ = exp − ωρσ S ρσ ψ. γ 3 ]. γµ ψ −→ Sγµ S −1 . (γ5 )S.The Dirac equation 31 Different representations can be related to each other by unitary transformations.46) (4.9 and 4.44) and an adjoint spinor defined by ψ = ψ † γ0 . (γµ )S. while L− 1γ µ L = Λµ γ ν .R. Another matrix which is often used is γ5 defined as γ5 = i γ 0 γ 1 γ 2 γ 3 = −i γ0 γ1 γ2 γ3 = It satisfies {γ5 . 2 (4.48) are projection operators.  (4. the rotation and boost generators in Weyl representation in Eqs 4. γ 2 ]. The ν 2 latter assures Lorentz invariance of the Dirac equation (see Exercise 4. that project out chiral right/left states.43) S=√  . is   1  1 1    (4.R. −→ Sψ. We note that ψ → Lψ and ψ → ψ L−1 . For instance in the Weyl representation (but valid generally). S −1 .47) (4.3) . γ µ } = 0 and explicitly one has    0 1  1  . = S(γµ )W. which in the case of the Weyl representation are just the upper and lower components.49) i with S ρσ = 1 σ ρσ = 4 [γ ρ . For example. Lorentz invariance The Lorentz transformations can also be written in terms of Dirac matrices.R. (4.

  ψ= (4. One has ψ c = −ψ T C −1 . The latter implies that the conjugate of a right-handed spinor. reverse the charge of the particle. ∂ (4. in Weyl representation we find in Dirac space     ∗ C  ξ    −→ ψ c = C ψ T =  −ǫ η  . that brings ψ → ψ c . (C)S. is a left-handed spinor. In S. (or W. (C)W. usually to be taken unity.58) where ηc is an arbitrary (unobservable) phase.51) The existence of positive and negative energy solutions implies another symmetry in the Dirac equation. =    0 ǫ iσ 2 (4.g. ∂ (4.55) (4. γ5 ] = 0.R.52)    η ξ one can apply space inversion. Some properties of C are C −1 = C † and C T = −C. e.) and all representations connected via a real (up to an overall phase) matrix S.g.6). x = (t. As with parity we look for a transformation. for instance. Note that we have (as expected) explicitly in Weyl x representation in Dirac space     P  ξ    −→ ψ p = γ 0 ψ =  η  .    0 −ǫ 0    0   −ǫ 0  .56) (i/ − M ) ψ c (x) = 0. This symmetry does not change the spin 1/2 nature.R.R.57) (4. x) → x = (t. Explicitly. = =   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. T (4. called charge conjugation. but it does. ψR . which is again a solution of the Dirac equation. such that T Cγµ C −1 = −γµ . 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.R.54) e.50) Charge conjugation but with ψ p (x) ≡ ηp γ0 ψ(˜) (Exercise 4. ∂ with the solution ψ c (x) = ηc Cψ (x).59)    η ǫ ξ∗ . In any representation a matrix C exist. parity.The Dirac equation 32 Parity There are a number of symmetries in the Dirac equation. C is real and one has C −1 = C † = C T = −C c and [C.53) (4.    −iσ 2   0 −ǫ  = .   ψ= (4. −x) and via a few manipulations obtain again the ˜ Dirac equation (i/ − M ) ψ p (x) = 0. (i/ − M ) ψ(x) = 0. T (4. It is easy to convince oneself that if ψ(x) is a solution of the Dirac equation.

¯ ′ ′     . s) + v(k. where u satisfies   −σ · p  Ep − M      u(p) = 0. 2M 2M (4.64) (4. the best representation to describe particles at rest is the standard representation. s) = u(p.63) u(p.The Dirac equation 33 4.70) =  Ps =  ˆ 0 1∓σ·s 2 2 . ψ(x) = u±s (p) e−i p·x .. s)¯(p. be obtained by a Lorentz transformation.68) (4. v (p. s ) = v (p.62) Explicit solutions in the standard representation are   χs       . s)v(p. s) and C v T (p. s′ ) = −2M δss′ .61)     σ·p −(Ep + M ) There are also two negative energy solutions. In an arbitrary frame one has s · p = 0 and s(p) can e. in which γ 0 is diagonal (see discussion of negative energy states in section 4. s)v(p. In that frame is a spacelike unit vector sµ = (0. with E = Ep = + p2 + M 2 . (2π)3 2Ek (4. p     −σ · p (Ep − M )     Ep + M     σ ·p ¯ Ep +M χs χs ¯ For a free massive particle.69) = − s In order to project out spin states. s)u(p. s).1). s) e−i k·x b(k. 2M 2M −/ + M p v(p.   ⇔ (/ − M ) u(p) = 0.60)       ∂ −iσ · ∇ −i ∂t − M (4.   ⇔ (/ + M ) v(p) = 0. s). ¯ (4. s) v = .g. s) ei k·x d∗ (k. Choosing χ = −ǫχ∗ (the equivalent spin ¯ 1/2 conjugate representation).65) (4. s) = Ep + M      σ ·p   χs Ep +M where χs are two independent (s = ±) two-component spinors.66) An arbitrary spin 1/2 field can be expanded in the independent solutions. s)v(p. where v satisfies   σ·p  −(Ep + M )      v(p) = 0.4 Plane wave solutions Looking for positive energy solutions ∝ exp(−iEt) one finds two solutions. s)u(p. s)¯(p. s) = v(p. s) u = . s′ ) = 2M δss′ . s′ ) = v † (p.    (4. Note that the spinors in the negative energy modes (antiparticles) could be two different spinors. p (4. ¯ ¯ u† (p. s)u(p. The explicit Dirac equation in the standard representation reads  ∂  iσ · ∇   i ∂t − M       ψ(x) = 0. ψ(x) = v ±s (p) ei p·x . the spinors satisfy C uT (p.67) It is straightforward to find projection operators for the positive and negative energy states P+ P− = s p /+M u(p. the spin polarization vector in the rest frame is the starting ˆ point. It is easy to check that   1  1±σ·s 1 ± γ5 / s ˆ 0    (4. s′ ) = 2Ep δss′ . v(p. s) . After separating positive and negative energy solutions as done in the case of the scalar field one has ψ(x) = s d3 k u(k. s) = u(p. s ) = 0. The ¯ ¯ solutions are normalized to u(p. (4.

while the negative helicity (λ = −1/2) is in essence lefthanded. +1/2) = √ 3 > > > > p − i p2 > .73) Also for helicity states C uT (p. Explicit examples of spinors are useful to illustrate spin eigenstates. λ = +1/2) = √ > > > > + E + M − √E − M > . > > > > . chirality. (4. λ). . one has in standard representation: 9 9 8 8 0 > > E+M > > > > > > > > > E+M > > > > > 1 1 0 >. > > > p E +M > E+M > > > > . One finds for the helicity states in Weyl representation: √ 9 9 8 √ 8 0 > > E+M + E−M > > √ > > > > > > > E + M − √E − M > > > 1 > 1 > 0√ > . : : 0 E +M + E −M √ 9 8 8 √ 9 0√ > > √ > − E+M − E−M > > > > > > > > > 1 > 1 > E+M − E−M > 0 > .48 replace the spin projection operators which are not defined (by lack of a rest frame). > > > > u(p. p ψR/L ≡ PR/L ψ are independent solutions. By writing the helicity states in Weyl representation it is easy to project out righthanded (upper components) and lefthanded (lower components). : : 0 − E −M 9 8 √ 8 > > − E−M > √ 0 > > > > > − E −M > > > > 0 >. : : . > > > > 0 √ > > 2> √ 2> . −1/2) = √ > E + M > −(E + M ) > : 0 9 > > > > >. This means that in the solution space for massless fermions the chirality states. The chirality projection operators in Eq. 4. λ = −1/2) = > > > > > > > E−M > > √ 0 .71) u(p. λ = +1/2) = > > > > > > − E+M > 0 > > > √ . helicity states. We note that for high energy ¯ ¯ the positive helicity (λ = +1/2) is in essence righthanded. etc. λ = +1/2) = > u(p.The Dirac equation 34 (the last equality in the restframe and in standard representation) projects out spin ±1/2 states (check ˆ ˆ this in the restframe for s = z ). v(p. : 1 : p + i p2 −p3 8 1 2 > p − ip > > > 1 −p3 > > v(p. For instance with the z-axis as spin quantization axis. = √ =⇒ 2 2E E+M + E−M ( E + M + E − M) which vanishes in the ultra-relativistic limit E ≫ M or in the massless case. −1/2) = √ (4. − E +M − E −M 0 9 > > > > >. > √ 0 > > > u(p. λ = −1/2) = > √ v(p. > > > v(p. 9 > > > > >. (4. > > > > .72) ˆ Helicity states (p along z ) in Standard representation are: 9 8 √ 8 > E+M > > √ 0 > > > > > > > > > E +M >. 8 −p3 > > > > −p1 − i p2 1 > > v(p. Note that for solutions of the massless Dirac equation pψ = 0. λ = −1/2) = √ > √ > > > > > > v(p. > > > > > > 1 u(p. > > > > . In principle massless fermions can be described by twocomponent spinors. λ) and C uT (p.74) . +1/2) = √ > 0 E +M > > : E+M 9 > > > > >. The ratio of the components is (directly or inversely) proportional to √ √ E+M − E−M 2M M ≪E M √ √ √ . > > > > . Therefore. γ5 /ψ = 0 but also / p p /γ5 ψ = −γ5 /ψ = 0. . λ) = v(p. λ) = u(p. λ = +1/2) = √ > √ > > E+M − E−M > > > > > 0√ > > > 2> 2> √ . u(p. λ = −1/2) = √ > >. : : 0 E+M (4.

(8) σµν ≡ (i/2)[γµ .s which appears in quantum electrodynamics calculations (see Chapter 10) for the emission of a photon from an electron changing its momentum from k to k ′ (k 2 = k ′2 = m2 ). s)¯k (k.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. namely the calculation of the quantity 1 ∗ (¯(k. (9) Tr(γ µ1 · · · γ µ2n ) = Tr(γ µ2n · · · γ µ1 ). (7) Tr(γ5 γ µ γ ν ) = 0 or Tr(γ5 //) = 0. s′ )¯i (k ′ .76) ul (k ′ . In principle one can take a representation and just calculate the quantity u(k.g µρ g νσ + g µσ g νρ ). s)γ ν u(k ′ . s′ )γ µ u(k. s′ )) = u(k ′ . (Use (γ5 )2 = 1 and pull one γ5 through.s′ ∗ (4. s′ )) (¯(k. (6) Tr(γ5 ) = 0. k k 2 .) In order to show the use of the above relations. (7) γµ γ5 = -γ5 γµ = i ǫµνρσ γ ν γ ρ γ σ /3!. It is. ab (4) Tr(γ µ γ ν γ ρ γ σ ) = 4 S µνρσ . s)γ µ u(k ′ . For this we first have to prove (do this) that (¯(k.s′ 1 ′ (/ + m)li (γ µ )ij (/ + m)jk (γ ν )kl k k 2 1 Tr [(/ ′ + m)γ µ (/ + m)γ ν ] . s)(γ ν )kl u u s. etc. s′ )(γ µ )ij uj (k. (10) γ µ γ ν = g µν − i σ µν (11) γ µ γ ν γ ρ = S µνρσ γσ + i ǫµνρσ γσ γ5 with S µνρσ = (g µν g ρσ .5 γ gymnastics and applications An overview of properties can be found in many text books. k ′ ) = 2 ′ s. we will give one example. s)γ µ u(k ′ . (6) γ5 = i γ 0 γ 1 γ 2 γ 3 = −i ǫµνρσ γ µ γ ν γ ρ γ σ /4!.The Dirac equation 35 4. (5) Tr(γ µ1 · · · γ µn ) = g µ1 µ2 Tr(γ µ3 · · · γ µn ) . s′ )) . s′ ) ¯ u s. γν ] (9) σµν γ5 = γ5 σµν = ǫµνρσ γ ρ γ σ /2. ab (8) Tr(γ5 γ µ γ ν γ ρ γ σ ) = −4i ǫµνρσ . u u (4.75) Lµν (k. s′ ). (Use matrices C to transpose the matrices in the expression under the trace and use Tr AT = Tr A. s)(γ ν )kl ul (k ′ . s)¯k (k. (5) γ µ γ λ γ ν γ ρ γ σ γµ = 2 γ σ γ λ γ ν γ ρ + γ ρ γ ν γ λ γ σ .) (2) Tr(1) = 4. s′ )(γ µ )ij uj (k. ab (4) γ µ γ ν γ ρ γ σ γµ = −2 γ σ γ ρ γ ν . 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). more convenient ¯ to use the projection operators introduced earlier. Properties of traces of γ-matrices (1) Tr(γ µ1 · · · γ µn ) = 0 if n is odd. however. u ¯ This leads with explicit Dirac indices to Lµν = = = = 1 2 1 2 ui (k ′ . (3) Tr(γ µ γ ν ) = 4 g µν or Tr(//) = 4 a · b. s)γ µ u(k ′ .

The Dirac equation

36

The trace is linear and can be split up in parts containing traces with up to four gamma-matrices, of which only the traces of four and two gamma-matrices are nonzero. 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 .

(4.77)

Exercises
Excercise 4.1
Prove Eq. 4.13, H(p) = exp φ·σ 2 = M +E+σ·p 2M (E + M ) ,

ˆ ˆ where E = M γ = M cosh(φ) and p = M βγ n = M sinh(φ) n. Do this e.g. by proving that H 2 (p) = exp (φ · σ) = (E + σ · p)/M = σ µ pµ /M. ˜

Exercise 4.2
Show that j µ = ψγ µ ψ is a conserved current if ψ en ψ are solutions of the Dirac equation

Exercise 4.3
Imposing Lorentz invariance on the Dirac equation can be achieved by requiring L−1 γ µ L = Λµ γ ν . ν µ µ Look at the infinitesimal forms Λµ = gν + ων and L = 1 − (i/2)ωρσ S ρσ to obtain ν [γ µ , i S ρσ ] = g ρµ γ σ − g σµ γ ρ , σ ρσ 1 . −i S ρσ = [γ ρ , γ σ ] = −i 4 2

Exercise 4.4
Show that in P+ ≡
s=±1/2

p /+M u(p, s)¯(p, s) u = 2M 2M

both sides are projection operators. Show that the equality holds, e.g. by letting them act on the four basis spinors u(p, s) and v(p, s).

Exercise 4.5
Show starting from the relation {γµ , γν } = 2gµν that γµ / γ µ a Tr(γ
µ1

=

Tr(//) = ab µn ···γ ) =

−2 / , a

4 a · b, 0 if n is odd.

Excercise 4.6
Apply space-inversion, x → x, to the Dirac equation and use this to show that the spinor ψ p (x) = ˜ γ0 ψ(˜), where x = (t, −x) is also a solution of the Dirac equation. x ˜

The Dirac equation

37

Exercise 4.7
(a) Show that (in standard representation)  j0 (kr)χm    ψ(x, t) = N     k ˆ i E+M σ · r j1 (kr)χm    −i Et   e ,   

is a solution of the (free) Dirac equation. In this solution the wave number k = |k| = and χm a two-component spinor while j0 and j1 are spherical Bessel functions.

√ E2 − M 2

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. (b) Next, we want to confine this solution to a spherical cavity with radius R. A condition for confining the fermion in the cavity is i / ψ = ψ. Show that this condition is sufficient to guarantee n that nµ j µ = 0, i.e. there is no current flowing through the surface of the cavity. Note: To prove this, you need to determine the corresponding confinement condition for ψ. The n conditions for ψ and ψ imply that nµ j µ = ψ/ ψ = 0. (c) Apply the confinement condition to find the lowest eigenmode in the spherical cavity for which ˆ nµ = (0, r). Calculate it for M R = 0, M R = 1.5 and M → ∞. (d) Plot for these cases the vector densities ψ † ψ and the scalar densities ψψ as a function of r/R.

Excercise 4.8
A negatively charged pion (π − ) decays in a lepton ℓ (ℓ can be an electron or a muon) and a corresponding anti-neutrino. The pion has spin 0, the lepton and neutrino have spin 1/2. The transition amplitude is given by p M = ℓ(kℓ )¯ℓ (kν )|Hint |π(p) = GF fπ u(kℓ )/(1 − γ5 ) v(kν ). ν Calculate |M |2 , which is the most important term contributing to the decay width (inverse lifetime). Note: the outcome shows that the decay into muons is much more likely that the decay into electrons (mµ = 105.6 MeV, me = 0.511 MeV). Actually if the lepton would be massless, the decay would even be forbidden. Do you understand why. The calculation of the life-time is completed in chapter 11.

Chapter 5

Vector fields and Maxwell equations
5.1 Fields for spin 1

In section 3.2 we have discussed the vectors ǫλ (with λ = 0 and ±1) as basis states of the spin 1 representation. In the rest frame plane waves of the form ǫλ ei k·x , thus, can describe spin 1 waves. The covariant generalization is to describe them with Vµ (x) = ǫµ (k) e±ik·x . These are solutions of the Klein-Gordon equation with in addition a condition, (2 + M 2 )Vµ = 0, with ∂µ V µ = 0 (Lorentz condition). (5.2) (5.1)

The Lorentz condition implies that k µ ǫµ (k) = 0. In the restframe, where k = k0 = (M, 0), there are then indeed three independent possibilities, namely ǫµ (k0 ) = (0, ǫλ ). The vectorfield is therefore (λ) suited to describe spin 1. In an arbitrary frame ǫµ (k) with λ = 0, ±(1) are obtained by boosting the restframe vectors. A real vector field Vµ (x) then can be expanded in modes as Vµ (x) =
λ=0,± (λ)

d3 k ǫ(λ) (k) e−ik·x c(k, λ) + ǫ(λ)∗ (k) eik·x c∗ (k, λ) . µ (2π)3 2E µ

(5.3)

Since the vectors ǫµ (k) together with the momentum k µ form a complete set of four-vectors, one has ǫ(λ) (k) ǫ(λ)∗ (k) = −gµν + µ ν kµ kν . M2 (5.4)

λ=0,±

5.2

The electromagnetic field
F µν = = ∂ µ Aν − ∂ ν Aµ  1 −E 2  0 −E  1   E 0 −B 3   2  3  E B 0   3 E −B 2 B 1 Aµ = (φ, A), 38 (5.5)  −E    B2    1 , −B    0
3

With the electromagnetic field tensor

(5.6)

where Aµ is the four-vector potential

(5.7)

15 the d’Alembertian is replaced by 2 + M 2 . |k|). For electrodynamics one has furthermore the freedom of gauge transformations. transverse or Coulomb gauge). Taking the divergence of this equiation. −i. The gauge freedom can first of all be used to impose the Lorentz condition for the electromagnetic field. Furthermore. it does not eliminate the freedom of gauge transformations but they are now restricted to Aµ → Aµ + ∂µ χ with 2χ = 0.5. only two independent vectors remain ǫ(±) µ (k) = (0. in accordance with what we found in section 3. It states the absence of magnetic charges and Faraday’s law.12) (5.11) This equation is not affected by a gauge transformation. it will vanish at all times. The equations of motion for the fields Aµ become 2Aµ − ∂µ (∂ν Aν ) = jµ . Note that the incorporation of the Lorentz condition can also be done for the case of the massive vector field.14) of which the solutions give the Li´nard-Wiechert potentials. The equation in vacuum. 5. for M = 0. The Lorenz condition can actually be incorporated by changing the equations of motion to 2Aµ + (λ − 1)∂µ (∂ν Aν ) = jµ . As remarked above. ∂µ Aµ = 0. shows that the electromagnetic fields correspond to the case of a massless spin 1 field. Fµν −→ Fµν + (∂µ ∂ν χ − ∂ν ∂µ χ) = Fµν . there is still a gauge freedom left (Aµ → Aµ + ∂µ χ with 2χ = 0). one finds for a conserved current and λ = 0 2∂µ Aµ = 0. Therefore if k µ = (|k|. ∓1. ˜ x Note that via parity transformation one obtains another solution. (5.e. .15) (5.9) 1 ˜ where F µν ≡ − 2 ǫµνρσ Fρσ is the tensor where E → B and B → −E. if ∂µ Aµ = 0 for large times |t|. if in Eq. In that case one also has ∇·A = 0 or ki ǫi (k) = 0. 39 (5. One possible choice is the gauge A0 = 0 (5.16) (5. 0. 0).Vector fields and Maxwell equations and the (conserved) current j µ = (ρ.8) (5. (5.10) the electric and magnetic fields are unchanged. (5. while the fact µ C ∗ that the Minkowski space is real implies that Aµ (x) = Aµ (x) = Aµ (x).13) Hence. i.18) corresponding to the helicity states λ = ±1 of the massless photon field. the antisymmetry of F µν implies that ˜ ∂µ F µν = 0 or ∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ = 0. j) the Maxwell equations read ∂µ F µν = j ν . Under a gauge transformation Aµ −→ Aµ + ∂µ χ. AP (x) = Aµ (˜). in which case one has 2Aµ = jµ .17) (radiation. 0. (5. (5. moree over. 2Aµ = 0. Although the Lorenz condition is a constraint.

20) 1 2 (∇ × A) · ds = B · ds = where Φ is the flux through the solenoid. may look a bit akward.1: The Aharonov-Bohm experiment 5. ˆ B = 0. 2 2 2 BR2 ˆ BR2 y BR2 x A= φ= − . This occurs in the region outside the solenoid. h ¯ The phase difference between the two paths of the electron then are given by ∆δ = − e h ¯ e = h ¯ A · dr + e h ¯ A · dr = e h ¯ e h ¯ A · dr e Φ. Inside solenoid: Outside solenoid: B = B z.1. . F = e E + ev × B (5. The phase difference between the electrons travelling two different paths is given by δ = 2π a a ∆x d = −≈ . cos φ. 0) and φ = (− sin φ. This phase difference is actually observed. λ L − λ λ or ∆x = (L− λ/d)δ. but where E = B = 0.19) are gauge invariant.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. 2r 2r2 2r2 ˆ where r = (cos φ. nevertheless. .0 . The situation. This causes an interference pattern as a function of ∆x.Vector fields and Maxwell equations 40 L ∆x d a (infinitely) long solenoid Figure 5. 0). The explanation of the significance of the vector ˆ potential A for the phase of the electron wave function is found in minimal substitution ψ ∝ exp i p·r h ¯ −→ exp i (p − eA) · r h ¯ e = exp −i A · r exp h ¯ i p·r . where Aµ = 0. Nevertheless the significance of Aµ is shown in the Aharonov-Bohm experiment.0 . while the electromagnetic field Aµ is not. To be precise. 5. A= By Bx Br ˆ φ= − . In this experiment an observable phase difference is measured for electrons moving through field-free space (E = B = 0 but Aµ = 0). The energy density 1 (E 2 + B 2 ) = 0 2 (hence a vacuum). nothing more than a manisfestation of a nontrivial vacuum (a la the Dirac sea for fermions). h ¯ (5. . In the presence of an electromagnetic field Aµ an additional phase difference is observed. It is however. sin φ. but A = 0 (hence there is structure in the vacuum). illustrated in fig.

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. The difference here is that we are working in a space with a ’defect’ (the infinitely long solenoid). The AharonovBohm experiment shows a realization of this possibility. 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. In such a space loops with different winding number (i. it can be obtained from the situation A = 0 by a gauge transformation. This situation that A can be written as ∇χ is always true when ∇ × A = 0. giving no observable phase2 . Now back to the Aharonov-Bohm experiment. is multivalued as φ = 0 and φ = 2π are the same point in space.21) (φ being the azimuthal angle). 5. If this would be be happening in empty space one would be in trouble.e. h 1 The occurrence of nontrivial possibilities. i. however. 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. A = ∇χ with χ= BR2 φ 2 (5. Therefore the gauge transformation must be uniquely defined in the space one is working in. the relevant charge turns out to be 2e because the electrons appear in Cooper pairs . h φ=0 ¯ The function χ in Eq. The observable phase going around in general is ∆δ = e h ¯ A · dr = e h ¯ ∇χ · dr = e φ=2π χ| . minimizing the space occupied by B fields. organizing itself in tiny flux tubes (Abrikosov strings).Vector fields and Maxwell equations U(1) S (space with solenoid) 41 class 0 ei φ class 1 Π1(S) = Z class 2 Figure 5. Consider a superconductor. forming a simple (connected) space. nonobservable phases φ = 2π n.2: The topology of the space in the Aharonov-Bohm experiment. the number of times they go around the defect) cannot be continuously deformed into one another. Therefore space outside the solenoid doesn’t care that χ is multi-valued.21. has been employed by Dirac in constructing magnetic monopoles in electrodynamics 2 Actually. Another situation in which the topology of space can be used is in the case of superconductors.e. Below the critical temperature the magnetic field is squeezed out of the superconducting material. implying that it must be single-valued1 .

Vector fields and Maxwell equations 42 Exercises Exercise 5. ∇×B =j+ ∇×E+ ∂E . .1 Show that Eqs.9 explicitly give the equations ∇ · E = ρ.8 and 5. 5. ∂t ∂B = 0. ∂t ∇ · B = 0.

t1 (6.8) 43 . ˙ p= ∂L . recall classical mechanics with the action t2 S= t1 dt L(x.5) with ∆x(t) = δx(t) + x(t) δτ .3) (6. These equations can be obtained from a lagrangian using the action principle.1 Euler-Lagrange equations In the previous chapter we have seen the equations for scalar fields (Klein-Gordon equation). ˙ 2 (6. (6.Chapter 6 Classical lagrangian field theory 6. x) ˙ (6. d dt ∂L ∂x ˙ = ∂L . x) = K − V = ˙ 1 mx2 − V (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 . Dirac fields (Dirac equation) and massless vector fields (Maxwell equations) and corresponding to these fields conserved currents describing the probability and probability current.1) and as an example the lagrangian L(x.7) The quantity ∂L/∂ x plays a special role and is known as the canonical momentum. As an example.6) The first term leads to the Euler-Lagrange equations. thus t′ = t + δτ. ∂x (6.2) The principle of minimal action looks for a stationary action under variations in the coordinates and time. x′ (t) = x(t) + δx(t). ∂x ˙ (6. and the total change x′ (t′ ) = x(t) + ∆x(t). The requirement δS = 0 with fixed boundaries x(t1 ) = x1 and x(t2 ) = ˙ x2 leads to t2 (6.

t2 ) d3 x . . = ∂R (R3 . dσµ . but which in 3 dimensions includes conserved quantities related to rotations and boosts. which in classical mechanics vanishes at the boundary) does not play a role. .6 can be rewritten as δS = . x). x) ≡ p x − L . = φ(x) + δφ(x) (6.Classical lagrangian field theory 44 For the lagrangian specified above this leads directly to Newton’s law p = m¨ = −∂V /∂x. with The resulting variation of the action is δS = R ′ = xµ + δxµ . indicated with δF [φ]/δφ should pose no problems.9) which is done because the first term (multiplying ∆x. t1 (6. . . x) and an action t2 S[φ] = t1 dt L = dt d3 x L (φ(x). . (R3 . 6.15) d4 x′ L (φ′ . ˙ x The second term in Eq. In classical field theory one proceeds in complete analogy but using functions depending on space and time (classical fields.2.e. The hamiltonian H is defined by H(p.18) 1 Taking a functional derivative.17) Furthermore the variations δφ and δ∂µ φ contribute to δS. (6. x′ ) − d4 x L (φ. gives rise to a conserved quantity (at least if H is invariant!) Q=p dx dλ −H dt dλ .13) (6. i.11) λ=0 of which p and H are the simplest examples related to space and time translations. ∂µ φ(x).12) Here R indicates a space-time volume bounded by (R3 . ∂µ φ′ . In general any continuous transformation. t2 ).14) (6. + ∂L ∆x − H δτ ∂x ˙ t2 . t1 ) and R3 . ∂µ φ(x)).10) One sees that invariance under time translations requires that H(t1 ) = H(t2 ). Consider a lagrangian density L which depends on these functions. giving1 δS = = δL δL δ(∂µ φ) + δφ + δ(∂µ φ) δφ R δL δL δφ + d4 x − ∂µ δφ δ(∂µ φ) R d4 x dσµ L δxµ ∂R dσµ ∂R δL δφ + L δxµ . indicated as x′µ φ (x) or combined φ′ (x′ ) = φ(x) + ∆φ(x). .16) The change in variables x → x′ in the integration volume involves a surface variation of the form dσµ L δxµ . . ∂R Note for the specific choice of the surface for constant times t1 and t2 . ˙ (6. . ∂µ φ. . also indicated by ∂R (a more general volume in four-dimensional space-time with some boundary ∂R can also be considered). ∆x = (dx/dλ)λ=0 δλ and δt = (dt/dλ)λ=0 δλ. H is a conserved quantity. . λ=0 (6. − d3 x . δ(∂µ φ) (6. R (6. L (φ(x).t1 ) (6. Variations in the action can come from the coordinates or the fields. think for instance of a temperature or density distribution or of an electromagnetic field). We will come back to it in a bit more formal way in section 9. ∂µ φ(x)) = R d4 x L (φ(x). their derivatives and possibly on the position.

L = 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 = − φ ∂µ ∂ µ + M 2 φ.29) . δL δ(∂µ ψ) δL δψ one obtains immediately i / −M ∂ → i = − γ µψ 2 i → ∂ / ψ − M ψ. 1/2 and 1 fields By an appropriate choice of lagrangian density the equations of motion discussed in previous chapters for the scalar field (spin 0). 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 ∗ (6. One easily obtains (2 + M 2 )φ(x) = 0. leading to the Euler-Lagrange equations. (6. (2 + M 2 )φ∗ (x) = 0. ∂ (6.Classical lagrangian field theory 45 With for the situation of classical fields all variations of the fields and coordinates at the surface vanishing.21) leading to (6.25) (6.20) (6. which differ only by surface terms.27) (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). 2 (2 + M 2 )φ(x) = 0. (6. Using the variations in ψ (in the symmetric form).24) ∂µ ∂ + M µ 2 φ.19) δ(∂µ φ) δφ 6.26) The Dirac field The appropriate lagrangian from which to derive the equations of motion is L = i → i ← i ↔ ψ / ψ − M ψψ = ψ / ψ − ψ / ψ − M ψψ ∂ ∂ ∂ 2 2 2 = ψ (i/ − M ) ψ. (6. δL δL ∂µ = . which can be considered as the sum of the lagrangian densities for two real scalar fields φ1 and φ2 √ with φ = (φ1 + iφ2 )/ 2. the second term is irrelevant. the Dirac field (spin 1/2) and the vector field (spin 1) can be found.22) (6. The scalar field It is straightforward to derive the equations of motion for a real scalar field φ from the following lagrangian densities. = 2 ψ = 0.2 Lagrangians for spin 0.

(6. . .31 rewritten in terms of ξ and η is “ ” LM (Dirac) = −M ξ † η + η † ξ . 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.33) 2 ≡ (ψL ) = C ψL c T c Rewriting the Majorana mass term in Eq. = 2 αβ = −βα.and lefthanded ones. R L the Majorana lagrangian is actually almost of the same form as the Dirac mass term. and thus (β ∗ α)∗ = α∗ β = −βα∗ . (6. We note that c ψL There exists a different possibility to write down a mass term with only one kind of fields. ∂ ∂ 2 2 (6. The Majorana case is actually more general.35) L = Note actually that using the following relations for Majorana spinors. The mass term in the Dirac lagrangian 6.39) 2 This term is written down with η being an anticommuting Grassmann number for which (αβ)∗ = β ∗ α∗ = −α∗ β ∗ .Classical lagrangian field theory 46 and similarly from the variation with respect to ψ ψ i / +M ∂ ← = 0. > (6. The spinor to be used satisfies 9 8 9 8 ∗ > 0 > c > > > ⇒ Υ ≡ > −ǫ η > . In that case one finds trivially L = ↔ ↔ 1 1 ψR i / ψR + ψL i / ψL − M (ψR ψL + ψL ψR ). Vector field From the lagrangian density L 1 λ 1 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 .36) (6. it is possible to introduce ’real’ spinors or Majorana spinors that satisfies Υc = Υ.32) Since the kinetic term in L separates naturally in ξ and η.33 in terms of ψL and ψL one finds 8 > −ǫ η ∗ => : 0 9 > > = ΥR . (6. L = ↔ ´ 1 1` Υi / Υ − ∂ M ΥR ΥL + M ∗ ΥL ΥR .30) It is often useful to link to the two-component spinors ξ and η which we started with in chapter 4.31) showing e. or equivalently separate the field into right. 6. η η ∗ for which η0 = ǫ η0 . (6. (6. Υc = ΥL and Υc = ΥR .37) Note the factor 1/2 as compared to the Dirac lagrangian. which comes because we in essence use ’real’ spinors.38) (6.34) Υ = Υ and ΥL = ψL = : . The reasons for Grassmann variables will become clear in the next chapter. 4 2 ↔ 1` 1 c´ c ψL i / ψL − ∂ M ψL ψL + M ∗ ψL ψL . : . 2 2 (6. namely2 ” 1 “ LM (Majorana) = + M η † ǫη ∗ − M ∗ η T ǫη .g. that the lagrangian separates into two independent parts for M = 0.g. Peshkin and Schroeder).

these conserved quantities become operators. Returning to δS the surface term is rewritten to δS = R d4 x .40) 6. (6. As an example for the classical field. δ(∂µ φ) (6. implying ∂µ F µν + M 2 V ν = 0 and ∂µ V µ = 0. λ=0 (6. (6.42) Θµν = The variation δS thus can be expressed as δS with J µ (x) = − = δL ∂ ν φ − L g µν . δ(∂µ φ) (6.47) . These do not affect the dynamics and will give rise to conserved currents. one can consider the variations at the initial or final surface. consider U (1) transformations of the Dirac field proportional to the charge e. ∆Φ = (dφ/dλ)λ=0 δλ and δxµ = (dxµ /dλ)λ=0 δλ. discussed in the next chapter. In this case. we consider the second term in Eq. 6. t).Classical lagrangian field theory 47 where Fµν = ∂µ Vν − ∂ν Vµ .3 Symmetries and conserved (Noether) currents Next. the presence of a symmetry that leaves the lagrangian invariant. 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. t). . .45) Thus considering ∂R = σ2 . In particular when the lagrangian is invariant under symmetries.18. ∂µ J µ (x) = 0. + R δL δL ∆φ − ∂ ν φ − L g µν δxν δ(∂µ φ) δ(∂µ φ) δL ∆φ − Θµν (x)δxν . which is the space-integral over the zero-component of a conserved current. δ(∂µ φ) For continuous transformations. Q(t1 ) = Q(t2 ). ψ(x) → ei eΛ ψ(x) or ∆ψ(x) = i eΛ ψ(x).43) F (σ1 ) − F (σ2 ) = dσµ J µ (x) = ∂R R d4 x ∂µ J µ (x).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 . requires the presence of a conserved quantity. Q= d3 x J 0 (x. + ∂R dσµ dσµ ∂R ≡ where d4 x . we get J µ (x) = − δL δ(∂µ φ) dφ dλ + Θµν (x) λ=0 dxν dλ .41) (6. which in a consistent picture precisely generate the symmetries. .44) δL ∆φ + Θµν (x)δxν . (6. (6.σ1 with σ = (R3 . .

50) (6. Lorentz transformations In this case the transformation of the coordinates and fields are written as δxµ ∆φi (x) = = ω µν xν . ↔ (6. (ω µν antisymmetric) 1 ωρσ (−i S ρσ )i φj (x). (6.4 Space-time symmetries One kind of symmetries that leave the lagrangian invariant are the Poincar´ transformations. 6. e. In the next section we will see the quantity show up as the charge operator. j 2 (6. (6. H = d3 x Θ00 (x). omitting the parameter Λ jµ = − δL δ(∂µ ψ ) i 48 · (−i e ψ ) − i δL · (i e ψ i ) δ(∂µ ψ i ) (6. For quantized fields this will become the expressions of the hamiltonian and the momentum operators in terms of the fields.56) . ˙ Q = d3 x j 0 (x. Therefore there are four conserved currents Θµν (ν labeling the currents!) and four conserved quantities Pν = dσµ Θµν = d3 x Θ0ν . 1 ωρσ (aρσ )µ xν ν 2 (6.54) (6.55) Note that the coordinate transformations can be written in a form similar to that for the fields.57) For the fields the behavior under Lorentz transformations has been the subject of previous chapters. = 0.48) = e ψγ µ ψ. we obtain (since δxµ = 0). (6. δxµ = with (aρσ )µ ν = g ρµ g σν − g σµ g ρν . t) is the conserved charge (Q = 0). For the complex scalar field the U (1) transformations leave the lagrangian invariant and lead to the conserved current (see Exercise 6.53) These are the energy and momentum.3) j µ = i e φ∗ ∂µ φ.51) (6.49) These currents are conserved as discussed already in chapter 2. = −(∂µ φ)δxµ = −ǫµ ∂µ φ(x). The integral over the zero-component. Summarizing.g. Translations µ Under translations (generated by Pop ) we have δxµ ∆φ(x) δφ(x) = ǫµ . From Noether’s theorem one sees that the current Θµν ǫν is conserved.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.Classical lagrangian field theory From the lagrangian for the Dirac field. including e the space-time translations and the Lorentz transformations.

63) (6. 2 (6. Gauge transformations of the second kind or local gauge transformations are transformations of the type φ(x) → ei eΛ(x) φ(x).67) i.e.68) (6.59) (6.5 (Abelian) gauge theories φ(x) → ei eΛ 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 . is not invariant under local gauge transformations. e. ∂µ T µν = ∂µ Θµν .69) (6. The lagrangians which we have considered sofar are invariant under global gauge transformations and corresponding to this there exist a conserved Noether current.66) In section 6.3). (6. In some applications (specifically coupling to gravity) it is advantageous to have an equivalent current that is symmetric. (6. (6.g.61) A final note concernes the symmetry properties of Θµν . .3 we have seen global transformations or gauge transformations of the first kind.70) → ei eΛ(x) ∂µ φ(x) + i e ∂µ Λ(x) ei eΛ(x) φ(x).65) 6.6) T µν = T νµ . • Vector field Aµ : −i Sρσ = aρσ . Defining T µν = Θµν − ∂ρ Gρµν . independent of x. (6. however. M µρσ (6. γσ ] This result for the Dirac field has been discussed in chapter 3 (see also Exercise 4.58) (6. one has a tensor that satisfies (exercise 6. in which the U (1) transformation involves an angle e Λ.64) µσ ρ =T µρ σ x −T x .62) with Gρµν = (H ρµν + H µνρ + H νµρ )/2. the angle of the transformation depends on the space-time point x. • Dirac field ψ: −i Sρσ = −(1/4)[γρ . → e−i eΛ(x) φ∗ (x). (6. where H ρµν is the quantity appearing in M ρµν . In general this tensor is not symmetric. φ(x) φ∗ (x) ∂µ φ(x) → ei eΛ(x) φ(x). The current following from Noether’s theorem is Jµ = = = = − δL ∆φi + Θµν δxν δ(∂µ φi ) δL 1 ωρσ − (−i S ρσ )i φj + Θµρ xσ − Θµσ xρ j 2 δ(∂µ φi ) 1 ωρσ {H µρσ + Θµρ xσ − Θµσ xρ } 2 1 ωρσ M µρσ . (6. Any lagrangian containing derivatives.Classical lagrangian field theory 49 • Scalar field φ: −i Sρσ = 0.

∂ 2 4 (6.75) The field φ is used here in a general sense standing for any possible field. 1 L (φ.78) . D ∂ A (6. Dµ φ) − Fµν F µν . 4 (6. (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. As an example consider the Dirac lagrangian. −e j µ Aµ = −e ρφ + e j · A. i L = ψγ µ ∂µ ψ − (∂µ ψ)γ µ ψ − M ψψ. Dµ φ(x) ≡ (∂µ + i eAµ (x))φ(x). 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 . The equation of motion for the fermion follow from δL δ(∂µ ψ) δL δψ = = i − γ µ ψ. A solution is the one known as minimal substitution in which the derivative ∂µ is replaced by a covariant derivative Dµ which satisfies Dµ φ(x) → ei eΛ(x) Dµ φ(x). (i/ − M ) ψ = (i/ − e/ − M ) ψ = 0.Classical lagrangian field theory 50 where the last term spoils gauge invariance. 2 i ∂ / ψ − M ψ − e/ ψ A 2 giving the Dirac equation in an electromagnetic 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 .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).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 . (6. (6. Replacing derivatives by covariant derivatives and adding the (free) part for the massless vector fields to the original lagrangian therefore produces a gauge invariant lagrangian. ∂µ φ) =⇒ L (φ. For this purpose it is necessary to introduce a vector field Aµ .73) (6.72) (6.71) Thus the covariant derivative has the correct transformation behavior provided Aµ → Aµ − ∂µ Λ.

Exercise 6. and deduce what is the charge of the antiparticle as compared to a particle. give the equation which is satisfied by ψ c = Cψ . ∂µ j µ = 0 (Note that the addition of a total derivative to the lagrangian density does not modify the equations of motion). . [iγ µ (∂µ − i e Aµ ) − M ]ψ = 0.3 Show that the current jµ = i φ∗ ∂µ φ for a complex scalar field is connected to a U (1) transformation on the fields. ↔ Exercise 6. ∂µ F µν = j ν .Classical lagrangian field theory 51 For the photon the equations of motion follow from δL δ(∂µ Aν ) δL δAν = = −F µν . φ → eiΛ φ.1 (a) Show that the Klein-Gordon equation for the real scalar field can be derived from the lagrangian density 1 1 L = ∂µ φ∂ µ φ − M 2 φ2 . (6. where j µ = e ψγ µ ψ.79) Exercises Exercise 6. 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. 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. −eψγ ν ψ. i. ∂µ F µν = j ν . Exercise 6.4 Given the Dirac equation for a (negatively charged) spin-1/2 particle in an external electromagnetic T field.e. giving the Maxwell equation coupling to the electromagnetic current. (c) Show that the interaction term is invariant under gauge transformations only if the current jµ is conserved.2 (a) Show that the homogeneous Maxwell equations can be derived from the lagrangian density 1 L = − Fµν F µν .

Classical lagrangian field theory 52 Exercise 6. t) = Enr ψu (r. Hints: it may be useful to realize that Gµρσ is antisymmetric in first two indices. .4. finally note that it is sufficient to show the last equation for dσµ M µρσ . t). in order to show that Θρσ − Θσρ = ∂µ H µρσ .5 Show that the Dirac equation for an electron with charge −e in an external electromagnetic field Aµ = (A0 . 2M 2M with Enr ≡ E − M ≪ M . Hint ≡ −gs µe · B = −gs 2me Exercise 6.4) for a stationary solution in the non-relativistic (|Enr | ≪ M ) and weak-field (|A0 | ≪ M ) limits yields in the standard representation the following equation for the ’upper (two) components’ ψu . in the coupling of spin for a spin 1/2 particle in a magnetic field.6 (optional) Prove the properties of the tensor T µν in section 6. use ∂µ T µν = 0 and ∂µ M µρσ = 0. A) (see also exercise 6. 1 e (−i∇ + eA)2 + σ · B − eA0 ψu (r. Qe s · B. gs = 2. Comment: this is the derivation of the g-factor.

is the Poisson bracket [A. 1 qx (t) = 3 d3 x φ(x. q) = p q − L(q. q) also considered in the previous chapter. starting from the lagrangian L(q. H]P dt and dO = [O. ˙ L(q. q(q.2) H(p. however. p]P = 1. q) ≈ (dO/dλ)λ=0 δλ under symmetry transformations one has dO = [O. An immediate generalization for fields can be obtained by considering them as coordinates.11) found through Noether’s theorem. Quantization of the theory is achieved by imposing canonical commutation relations between q and p. Q]P . in our example p = mq. there are conserved ˙ ˙ quantities Q (Eq.5) ∆ x ∆3 x 53 . namely [q. (7. ˙ 2 (7. dq The basic Poisson bracket is actually the one between q and its canonical momentum p. [q. labeled by the position. 6. p]P . Actually the conserved quantities satisfy a similar algebra as the Lie algebras for continuous symmetries that they are obtained from. (7.1) The Hamiltonian (also corresponding to a conserved quantity because of time translation invariance) is given by p2 + V (q). (7.1 Canonical quantization We will first recall the example of classical mechanics for one coordinate q(t).3) For variations δO(p. The bilinear product for the conserved quantities. Further. t). dλ Examples are dO = [O. q) = ˙ 1 mq 2 − V (q). B]P = dB dA dA dB − .Chapter 7 Quantization of fields 7. p] = i. This is nothing else as representing the classical Poisson algebra as operators in the Hilbert space. dq dp dq dp (7.4) with a possible realization as operators in the Hilbert space of (coordinate space) wave functions through qop ψ(q) = qψ(q) and pop ψ(q) = −idψ/dq. p)) = ˙ ˙ 2m where the (canonical) momentum p = ∂L/∂ q.

11) (7. ∂µ φ(x)) = ˙ d3 x L (φ. ∂ qx ˙ ∂ qx ˙ (7. (7. Sometimes it is useful to keep in mind that. t)] = [Π(x. albeit with ’sharp’ functions. a) = Rj (Λ−1 ) φj (Λ−1 x − a). The following items are essential to consistently quantize a theory.8) (7. The lagrangian is given by L = = x In order to construct the hamiltonian it is necessary to find the canonical momenta. 2 2 2 = (7. (7. formally.14) In fact. discussed in the section 6.15) with furthermore the relations [φ(x. qx .10) x ∆3 x [Πx qx − L x ] ˙ ∆3 x Hx (7.7) d3 x L (φ(x). • Canonical commutation relations The quantization of the theory is achieved by imposing the canonical quantization condition [qx (t). φ(y. i U † (Λ. qx+∆x ) . we have L (x) 1 1 ∂µ φ∂ µ φ − M 2 φ2 2 2 1 1 1 (∂0 φ)2 − (∇φ)2 − M 2 φ2 = 2 2 2 δL = = ∂0 φ.2. t).9) Note that this indeed corresponds to the zero-zero component (Θ00 ) of the conserved energy-momentum stress tensor Θµν . py (t)] = i δxy or for the fields φ(x) and Π(x) the socalled equal time commutation relations [φ(x. As an example.6) ∆3 x L x (qx . the discretization procedure above is an explicit example. for symmetry transformations. ∇φ) (7.13) Π(x) H (x) For the quantization procedure we can formulate a number of basic axioms of quantum field theory. ˙ δ φ(x) (7. φ. Π(y.16) . δ∂0 φ) 1 1 1 = Θ00 (x) = (∂0 φ)2 + (∇φ)2 + M 2 φ2 . • Poincar´ invariance e The full variation of the fields.Quantization of fields 54 etc. for the scalar field theory. px (t) = ∂L ∂Lx = ∆3 x = ∆3 x Πx (t). the fields φ(x) can be considered as regular operators in the Hilbert space after smearing with a test function f . ∆φ.12) (7. t)] = 0. t). (7. The hamiltonian then is H= x px qx − L = ˙ = x where H (x) = δL ˙ φ(x) − L (x) = Θ00 (x). Π(y. This implies the following transformation properties in the Hilbert space for the quantum fields φ. Φ(f ) = d4 x φ(x) f (x). t)] = i δ 3 (x − y). ˙ where Πx (t) is obtained from the continuous field Π(x) ≡ δL /δ(∂0 φ). a) φi (x)U (Λ. t).

18) with S µν given in 6. 1 Q = √ (a + a† ) 2ω and P = −i ω (a − a† ) 2 (7. 2 2 (7.21) Writing Q and P in terms of creation (a† ) and annihilation (a) operators. 2 i ωµν (S µν )i . Φ(g)] = 0. P ] = i. j 2 one has [φi (x). P ] = 0 (7.25) . sometimes referred to as second quantization. a† ] = 1.24) It is straightforward to find the commutation relations between N and a and a† .Quantization of fields 55 or if U (Λ. and [N. P ] 2 (7. [N.20) where the coordinate Q and the momentum P satisfy the canonical commutation relations [Q. φ(y)] = 0 if (x − y)2 < 0. (7. [Q. a] = −a.17) (Sµν )i φj (x). The measurements cannot influence each other or [Φ(f ). ω a† a + 2 2 ω i − [Q. H = = ω ω P P Q−i √ Q+i√ 2 2 2ω 2ω 1 1 =ω N+ . [φ(x). a) = i Rj (Λ−1 ) = 1 + iǫµ P µ − 1− i ωµν M µν . (7.19) 7. Q] = [P. 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.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. (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.4. a] = [a† . Microscopic causality implies local commutativity. a† ] = a† . Pµ ] = [φ (x). [a. • Causality Operators Φ(f ) and Φ(g) for which the test functions are space-like separated can be measured simultaneously (macroscopic causality). In simplified form the hamiltonian is given by H= 1 2 1 2 2 P + ω Q . i(xµ ∂ν − xν ∂µ )φ (x) + i (7.23) The hamiltonian in this case can be expressed in the number operator N = a† a. Mµν ] = i i∂µ φi (x).

From the normalizations one obtains a |n = √ n + 1 |n + 1 and a|n = n |n − 1 .like’ commutation relations between a(k) and a† (k ′ ).29) It is straightforward to prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with ’harmonic oscillator . † √ a and a act as raising and lowering operators. d3 x φ(x) i ∂0 e−i k ·x ↔ ′ ↔ (7.g.3 The real scalar field We have expanded the (classical) field in plane wave solutions. [φ(x). but one needs to realize that in that case k 0 = Ek = k2 + M 2 . 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.33) . i. (n − 1) a|n † i. and (7.27) 2 (2π)3 It is easy to check that these equations can be inverted (see Exercise 2. which represents the ’number of particles’ with momentum k. Π(x )]x0 =x′0 = 0. the real scalar field φ(x) becomes d3 k (7. φ(x) = (2π)3 2Ek where the Fourier coefficients a(k) and a† (k) are now operators. e.4 for the classical field) a(k) a† (k ′ ) = = d3 x ei k·x i ∂0 φ(x). φ(x )]x0 =x′0 = [Π(x). which we have split into positive and negative energy pieces with (complex) coefficients a(k) and a∗ (k) multiplying them. N |n = n|n one finds N a† |n N a|n = = (n + 1) a† |n . The quantization of the field is achieved by quantizing the coefficients in the Fourier expansion. 7. (2π)3 2Ek (7. a(k ′ )] = [a† (k). The canonical momentum becomes −i d3 k ˙ Π(x) = φ(x) = a(k) e−i k·x − a† (k) ei k·x . is equivalent with [a(k).e. H = = = d3 x 1 1 1 (∂0 φ)2 + (∇φ)2 + M 2 φ2 2 2 2 d3 k Ek † a (k)a(k) + a(k)a† (k) (2π)3 2Ek 2 d3 k Ek N (k) + Evac . (7. a† (k ′ )] = 0. Π(x′ )]x0 =x′0 = i δ 3 (x − x′ ) ′ ′ and (7.31) The hamiltonian can be rewritten in terms of a number operator N (k) = N (k) = a† (k)a(k). a† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ) [a(k).26) a(k) e−i k·x + a† (k) ei k·x .32) (7.28) (7.e.30) [φ(x).Quantization of fields 56 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. Note that we will often write a(k) or a† (k).

. (2π)3 2Ek d3 k a† (k) ei k·x . which now contains an infinite number of oscillators. and multiparticle states |(k1 )n1 (k2 )n2 .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).39) The problem with the vacuum or zero-point energy. |k = a† (k)|0 .38) d4 k 2π δ(k 2 − M 2 ) θ(k 0 ) |k k| = (2π)4 (7.Quantization of fields 57 where the necessity to commute a(k)a† (k) (as in the case of the quantum mechanics case) leads to a zero-point energy. This term will be adressed below. k 0 = Ek = k2 + M 2 ). 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). in field theory also referred to as vacuum energy Evac = 1 V 2 d3 k Ek . .45) (7.37) (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 √ √ . 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. n1 ! n2 ! (7. defining particle states |k = |k (with positive energy. (2π)3 2Ek (7.e. assuring that the vacuum (by definition) has eigenvalue 0!. d3 k |k k|. The rest of the states are then obtained from the groundstate via the creation operator. . For the purpose of normal ordering it is convenient to decompose the field in positive and negative frequency parts. |0 . (2π)3 2Ek (7.35) where the vacuum contribution disappears because of rotational symmetry.44) p|φ− (x)|0 = ei p·x . writing all annihilation operators to the right of the creation operators. (2π)3 2Ek (7.41) (7. which amounts to redefining H as H= d3 x : H (x) := d3 k Ek N (k).42) (7.46) (7. . d3 k a(k) e−i k·x . i.34) where V = (2π)3 δ 3 (0) is the space-volume. (2π)3 2Ek (7. .40) This procedure is known as normal ordering.36) and satisfying the completeness condition 1 = k|k ′ = (2π)3 2Ek δ 3 (k − k′ ). is solved by subtracting it as an (infinite) constant. (2π)3 (7. φ(x) φ+ (x) φ− (x) = = = φ+ (x) + φ− (x). a(k)|0 = 0.

50) 3 ′ [φ− (x).48) (7.54) = i∆(x − y) = i (∆+ (x − y) + ∆− (x − y)) . a† (k ′ )] ′ e (2π)3 2Ek (7. (iv) The equal time commutation relations follow from ∂ ∆(x) ∂t (v) For M = 0. (ii) ∆(x) is a solution of the homogeneous Klein-Gordon equation.e.Quantization of fields 58 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. ∆(x) = − ǫ(x0 ) δ(x2 ).18).49) (7. 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). (2π)4 d4 k θ(k 0 ) 2π δ(k 2 − M 2 ) ei k·x − (2π)4 d4 k − θ(−k 0 ) 2π δ(k 2 − M 2 ) e−i k·x = −i∆+ (−x) (2π)4 (7. (2π)4 (7. φ(y)] consider µν [φ+ (x). (2π)3 2Ek ≡ i∆− (x − y). (7.56) = −δ 3 (x). φ+ (y)] d3 k e−i k·(x−y) . φ− (y)] ≡ i∆+ (x − y). d3 k = − ei k·(x−y) = −i∆+ (y − x).51) (7. d3 k = (2π)3 2Ek = (7.47) d k −i k·x+i k′ ·y [a(k). x) = 0 and hence ∆(x) = 0 for x2 < 0. (7. i. 2π (7.17) and (7.55) d4 k ǫ(k 0 ) 2π δ(k 2 − M 2 ) e−i k·x . φ(y)] which has the following properties (i) i∆(x) = i∆+ (x) + i∆− (x) can be expressed as i∆(x) = where ǫ(k 0 ) = θ(k 0 ) − θ(−k 0 ).53) t=0 . The last item to be checked for the scalar field are the causality condition.52) The result for the invariant commutator function is [φ(x). (2π)3 2Ek or as integrals over d4 k. (iii) ∆(0. that they satisfy the required commutation relations in Eqs (7.

60) = i φ∗ ∂µ φ.e.58) and satisfy the equal time commutation relation (only nonzero ones) [φ(x). b† (k ′ )] = (2π)3 2Ek δ 3 (k − k′ ).64) The commutator of φ and φ† is as for the real field given by [φ(x).61) (7. ∂0 φ† (y)]x0 =y0 = i δ 3 (x − y). which is equivalent to the relations (only nonzero ones) [a(k). (2π)3 2Ek (7.57) (7.4 The complex scalar field In spite of the similarity with the case of the real field. extending it with the charge operator and the introduction of particle and antiparticle operators. (2π)3 2Ek d3 k b(k) e−i k·x + a† (k) ei k·x . The field satisfies the Klein-Gordon equation and the density current (U (1) transformations) and the energy-momentum tensor are jµ Θµν The quantized fields are written as φ(x) = φ† (x) = d3 k a(k) e−i k·x + b† (k) ei k·x . 2 From the lagrangian density (7.66) . φ† (y)] = i∆(x − y). we will consider it as a repetition of the quantization procedure. a† (k ′ )] = [b(k).5 The Dirac field L = ↔ i ψγ µ ∂µ ψ − M ψψ.59) (7. The hamiltonian is as before given by the normal ordered expression H = = = d3 x : Θ00 (x) : d3 k Ek : a† (k)a(k) + b(k)b† (k) : (2π)3 2Ek d3 k Ek a† (k)a(k) + b† (k)b(k) . ↔ (7.62) (7.65) 7. Also the charge operator requires normal ordering (in order to give the vacuum eigenvalue zero). (2π)3 2Ek (7.Quantization of fields 59 7. particles (created by a† ) and antiparticles (created by b† ) with the same momentum contribute equally to the energy. 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) . (7.63) i. (2π)3 2Ek (7. = ∂{µ φ∗ ∂ν} φ − L gµν .

73) (7. (2π)3 2Ek (7. (2π)3 2Ek (7.79) Also for the field and the canonical conjugate momentum anticommutation relations are considered. s)¯(k. s)u† (k.77) Note that achieving normal ordering. s)vj (k. (2π)3 2Ek d3 k b† (k.72) ψ(x) = s In terms of the operators for the b and d quanta the hamiltonian and charge operators are (omitting mostly the spin summation in the rest of this section) H = = Q = = d3 x : ψ † (x) i∂0 ψ(x) : d3 k Ek : b† (k)b(k) − d(k)d† (k) : (2π)3 2Ek d3 x : ψ † ψ : d3 k : b† (k)b(k) + d(k)d† (k) : . {b(k. s)u(k.70) where the last line is obtained by using the Dirac equation.67) i ↔ ψ / ψ − M ψψ gµν . ψj (y)}x0 =y0 = d3 k (2π)3 2Ek ui (k.71) (7. The solution is the introduction of anticommutation relations.78) (7. s).74) (7. (7. (7. s′ )} = {d(k. s)ei k·x + d(k. s) e−i k·(x−y) j s † + vi (k.69) (7. ∂ 2 (7. d† (k ′ . ˙ δ ψ(x) ↔ i Θ00 (x) = − ψγ i ∂i ψ + M ψψ 2 = iψγ 0 ∂0 ψ = iψ † ∂0 ψ. 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) . jµ Θµν = = ψγµ ψ. u v (2π)3 2Ek (7. s′ )} = (2π)3 2Ek δ 3 (k − k′ ) δss′ . s)e−i k·x . The quantized fields are written ψ(x) = s d3 k b(k.68) The canonical momentum and the hamiltonian are given by Π(x) H (x) = = δL = i ψ † (x).76) which seems to cause problems as the antiparticles (d-quanta) contribute negatively to the energy and the charges of particles (b-quanta) and antiparticles (d-quanta) are the same. s)e−i k·x + d† (k. s)v(k. i.75) (7. interchanging creation and annihilation operators. s)ei k·x .Quantization of fields 60 the conserved density and energy-momentum currents are easily obtained. † {ψi (x). s)¯(k.e. ↔ i ψγµ ∂ν ψ − 2 (7. s). . s) ei k·(x−y) x0 =y 0 . b† (k ′ .

4 2 This gives the equations of motion discussed before. −λ(∂ρ Aρ ). It is possible to continue in a covariant way with the lagrangian 1 λ L = − Fµν F µν − (∂µ Aµ )2 . ∂ (7. ∂ (7. ψj (y)}x0 =y0 = d3 k (/ + M )il (γ0 )lj e−i k·(x−y) k (2π)3 2Ek + (/ − M )il (γ0 )lj ei k·(x−y) k x0 =y 0 = d k 2Ek (γ0 γ0 )ij ei k·(x−y ) (2π)3 2Ek (7.83) From the lagrangian density the canonical momenta are = = (7. When we would have started with commutation relations for the field ψ and the canonical momentum. 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). one has † {ψi (x). ˙ δ A0 δL = F i0 = E i . ψ j (y)} at arbitrary times one has {ψi (x). (7.Quantization of fields 61 Using the positive and negative energy projection operators discussed in section 4. 4 Π0 Πi δL = 0.82) where i∆1 = i∆+ −i∆− .87) If one wants to impose canonical commutation relations ∂µ Aµ = 0 cannot hold as an operator identity. as the vanishing of Π0 induces a constraint.81) where i∆ = i∆+ + i∆− is the same invariant commutator function as encountered before.84) (7.86) (7. we would have obtained [ψi (x).88) . ˙ δ Ai (7. it implies the Lorentz constraint and leads to the canonical momenta Π0 Π i = = E. but we must restrict ourselves to the weaker condition B|∂µ Aµ |A = 0. ψ j (y)] = (i/x + M )ij i∆1 (x − y).6 The electromagnetic field 1 L = − Fµν F µν .80) 3 = δ 3 (x − y) δij . i (7. but has the problem of being noncovariant. 7.85) which reflects the gauge freedom. 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). For the scalar combination {ψi (x).

pj ]P .Quantization of fields 62 for physical states |A and |B . j = 1. k 3 ) this reads |k 3 |c(0) (k) − k 3 c(3) (k) |A = 0 c(0) (k) ∓ c(3) (k) |A = 0. The quantized field is expanded as Aµ (x) = d3 k (2π)3 2Ek (λ) 3 ǫ(λ) (k) c(k. The hamiltonian in terms of the creation and annihilation operators is (after normal ordering) given by H= d3 k Ek (2π)3 2Ek 3 λ=1 c† (k. which reproduce the Hamilton equations.95) Exercises Exercise 7. c† (k ′ . for which it is sufficient that ∂µ Aµ |A = 0. (d) Use Poisson brackets to calculate q and p.94) Choosing k µ = (|k 3 |. λ)ei k·x .3) satisfy the conditions of a Lie algebra. Πν (y)] = 0.90) where Πν = F ν0 − λ g ν0 (∂ρ Aρ ) and we have furthermore [Aµ (x). Aν (y)] = [Πµ (x).92) Note that for the transverse states there are no problems with the normalization and the statistics of the states as −g ij = δ ij for i.11. 2. rj ]P . and are equivalent with [c(k. and [ℓi . 0. ˙ ˙ . 0)c(k. It gives 3 k µ ǫ(λ) (k)c(k. (7. where Q is the conserved quantity in Eq. q) that dO/dλ = [O.89) with four independent vectors ǫµ . (b) Proof for a quantity O(p. µ λ=0 (7. µ λ=0 (7. In fact the commutation relations imply ˙ [Aµ (x). λ). Q]P . (7. λ)e−i k·x + c† (k. 6. ′ (7. 0) . i. Aν (y)]x0 =y0 = i gµν δ 3 (x − y).91) (7. Πν (y)]x0 =y0 = i gµν δ 3 (x − y). λ)c(k. i.e. where Aµ + + is the part of the vector field containing the annihilation operators. what are [ri . one time-like photon by itself is not allowed! This solves the problems with the normalization and the negative energies. The canonical equal time commutation relations are [Aµ (x). pj ]P . a longitudinal photon and two transverse photons. 7.93) which does exhibit problems with the time-like photon. λ) − c† (k. These problems are solved by the Lorentz constraint between physical states given above. λ)|A = 0. 0. (7. (c) Calculate the Poisson brackets for the coordinates x and the standard conserved quantities p and ℓ. containing a time-like photon. [ℓi .e.1 (optional) If you haven’t seen Poisson brackets in classical mechanics you should (a) Proof that Poisson brackets (Eq. λ′ )] = −g λλ 2Ek (2π)3 δ 3 (k − k′ ).

(2 + M 2 ) ∆hom (x) = 0. (b) Check the above commutation relation [φ(x). ∆hom (Λx) = ∆hom (x) if f is an invariant function. . (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.3 (a) Show that requires [φ(x).4 (Green’s functions) (a) Show that ∆hom (x) = 1 (2π)4 d4 k e−ik·x δ(k 2 − M 2 )f (k) is a solution of the homogeneous Klein-Gordon equation. Pµ ] = i∂µ φ(x). (e) Show that ∆R (x) = 0 if x0 < 0. Is the subtraction of ’zero point’ contributions essential in this check? e−i a µ Pµ φ(x) ei a µ Pµ = φ(x − a) Exercise 7. It is invariant. (2 + M 2 ) ∆inhom (x) = −δ 4 (x).2 Prove that the equal time commutation relations between φ(x) and Π(x′ ) are equivalent with the commutation relations between a(k) and a† (k ′ ) Exercise 7. which amounts to shifting the poles into the lower complex plane. Give also the expression for ∆C .Quantization of fields 63 Exercise 7. f (Λk) = f (k) for Lorentz transformations Λ. (Similarly one can show that for the advanced Green’s function ∆A (x) = 0 if x0 > 0). (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): k -E 0 ∆ R (retarded) ∆ C (causal) E ∆ ∆A (advanced) Show that for ∆R we can write ∆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.

f (k) ∝ θ(k 0 ). (g) Show that the homogeneous solution ∆ in (f) satisfies ∆(0. 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. (h) Show that the causal Green’s function ∆C (x) = θ(x0 ) ∆+ (x) − θ(−x0 ) ∆− (x).Quantization of fields 64 (f) By performing the k0 integration over a closed path C going around the poles show that ∆(x) = − 1 (2π)4 d4 k e−ikx C 1 k2 − M 2 are homogeneous solutions (of the form in a) for the following contours: k0 ∆+ ∆_ E ∆ -E (Note that one needs actual discontinuous functions such as the step function. (i) (Optional) Give similar expressions for the other Green’s functions under (d) in terms of the homogeneous solutions under (f). .

1 Parity xµ = (t.Chapter 8 Discrete symmetries In this chapter we discuss the discrete symmetries.1: The behavior of classical quantities under P.2) where ηP is the intrinsic parity of the field and A is a 4 × 4 matrix acting in the spinor space. Both ψ p and ψ satisfy the Dirac equation. ˜ (8. parity (P). x (8. T and C for classical quantities is shown in the table 1. time reversal (T) and charge conjugation (C). starting with the Dirac equation for ψ(x). 8. Table 8. The consequences of P. writing −1 ψ(x) −→ Pop ψ(x)Pop = ηP Aψ(˜) ≡ ψ p (x). and C quantity t r xµ E p pµ L s ˆ λ = s·p P t -r xµ ≡ xµ ˜ E -p pµ ˜ L s −λ T -t r −˜µ x E -p pµ ˜ -L -s λ C t r xµ E p pµ L s λ 65 .1) The parity operator transforms We will consider the transformation properties for a fermion field ψ(x). We can determine A. −r). (iγ µ ∂µ − M ) ψ(x) = 0. r) −→ xµ ≡ xµ = (t. T.

m). λ)u(k. −λ) e−ik·x − d† (k. −λ) eik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηP b(k. m) = = ˜ u(k. 8. λ)v(k. x µ (iγ ∂µ − M ) γ0 ψ(˜) = 0.Discrete symmetries 66 After parity transforming x to x the Dirac equation becomes after some manipulations ˜ ˜ x iγ µ ∂µ − M ψ(˜) = 0. λ) e−ik·˜ + d† (k. ˜ Pop d(k. ˜ −v(k. . −λ). op P (8. λ) Pop = ηP b(k. λ)u(k. For the scalar field we have seen −1 φ(x) −→ Pop φ(x) Pop = ηP φ(˜).11) (8.7) = λ d3 k x x ηP b(k. x Therefore γ0 ψ(˜) is again a solution of the Dirac equation and we have x ψ p (x) = γ0 ψ(˜). if helicity λ is used instead of the z-component of the spin m. the above operation reverses the sign of λ). choosing ηP is real (ηP = ±1) particle and antiparticle have opposite parity.5) (8. γ x (8. −λ) e−ik·x − d† (k. m). x The latter behavior of the vector field will be discussed further below. 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)γ0 v(k. λ) e−ik·x − d† (k.10) i.8) = λ = λ = λ From this one sees immediately that −1 ˜ Pop b(k. λ) eik·x . x (8. −λ)u(k. (2π)3 2Ek (8. λ) eik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηP b(k. iγ µ† ∂µ − M ψ(˜) = 0.9) (8.13) ψ(x) −→ ψ c (x) = ηC C ψ (x).4) (8. λ)v(k.6) (check this for the standard representation. λ)γ0 u(k. −λ)v(k.12) and for the vector field −1 Aµ (x) −→ Pop Aµ (x) Pop = −Aµ (˜). In the same way as the Fermion field. −λ) eik·x 3 2E (2π) ˜ k d3 k ˜ ˜ ηP b(k.e. −λ).2 Charge conjugation We have already seen the particle-antiparticle symmetry with under what we will call charge conjugation the behavior T (8. one can also consider the scalar field and vector fields. The result is ψ p (x) −1 = Pop ψ(x) Pop = ηP γ0 ψ(˜) x (8. m) γ0 v(k.3) (8. (i˜ µ ∂µ − M ) ψ(˜) = 0.

e. it is antilinear1 . m). to find ψ t (−˜) that is a solution of the Dirac equation. λ)C v T (k. λ) e−ik·x + b† (k. For a quantized field one has −1 ∗ −1 Top fk (x)bk Top = fk (x)Top bk Top . (8. λ) Cop = ηC d(k.Discrete symmetries 67 the latter being also a solution of the Dirac equation. we start with the complex conjugated x Dirac equation for ψ. λ) eik·x . 8. The same relations hold for helicity states. Therefore ψ c (x) = = λ −1 ¯ Cop ψ(x) Cop = ηC C ψ T (x) (8. The (time-reversed) Dirac equation becomes.16) d k ηC d(k. One has Aφ|Aψ = φ|ψ λ∗ A|φ µ∗ A|ψ ∗ = ψ|A† Aφ = ψ|φ . ˜ x −(iγ µ )∗ ∂µ − M ψ(−˜) = 0. r) −→ −˜µ ≡ −xµ = (−t.18) (8.20) The time reversal operator transforms We will again consider the transformation properties for a fermion field ψ(x). m) ¯ T = = v(k.14) (8. λ) e−ik·x + b† (k.21) where A is a 4 × 4 matrix acting in the spinor space.22) = Aψ|Aφ ∗ is antilinear if A(λ|φ + µ|ψ ) = + .3 Time reversal xµ = (t.15) (where one must be aware of the choice of spinors made in the expansion. x µ (iγ ∂µ − M ) γ5 Cψ(−˜) = 0. λ). Norm conservation requires Top to be anti-unitary2 . m). λ).19) Cop d(k. m) ¯ C v (k. γ x i(γ5 C)−1 γ µ γ5 C∂µ − M ψ(−˜) = 0.17) 3 = λ This shows that −1 Cop b(k. λ) −1 Cop = ∗ ηC b(k. x (8. x (8. The action on the spinors (using C = iγ 2 γ 0 = −iρ1 σ 2 in standard representation) gives C uT (k. x 1A 2 An −iC −1 γ µ C∂µ − M ψ(−˜) = 0. antilinear operator is anti-unitary if A† = A−1 . writing −1 ψ(x) −→ Top ψ(x)Top = ηT Aψ(−˜) ≡ ψ t (−˜). (2π)3 2Ek (8. As time reversal will transform ’bra’ into ’ket’. u(k. λ)u(k. ((iγ µ )∗ ∂µ − M ) ψ(x) = 0. i. r). Top |φ = φt | = (|φt )∗ . λ)C uT (k. λ)v(k. λ) eik·x ¯ ¯ (2π)3 2Ek d3 k ηC d(k. . x x (8. x iγ µT ∂µ − M ψ(−˜) = 0. (i˜ µ∗ ∂µ − M ) ψ(−˜) = 0. (8. as discussed in section 4).

p.24) (8. λ) T −1 = η ∗ d(k. λ Therefore γ5 Cψ(−˜) is again a solution of the (ordinary) Dirac equation and we can choose (phase is x convention) ψ t (x) = i γ5 C ψ(−˜). op T (8. p. λ| ¯ C |¯. λ)u∗ (k. λ) eik·˜ + d† (k. λ a |a. −λ a T a. They are the following S(x) V µ (x) T µν (x) Aµ (x) P (x) ¯ = ψ(x)ψ(x) (scalar) µ ¯ = ψ(x)γ ψ(x) (vector) µν ¯ = ψ(x)σ ψ(x) (tensor) ¯ = ψ(x)γ5 γ µ ψ(x) (axial vector) ¯ = iψ(x)γ5 ψ(x) (pseudoscalar) (8. λ) eik·x + d† (k. λ)u∗ (k. Since there are 16 independent 4 × 4 matrices. p. λ).34) . −λ |¯. λ a P |a.31) (8. λ) eik·x + d† (k. λ| a. λ)iγ5 Cv(k.25) (check this for the standard representation).32) (8. −p. λ)iγ5 Cu(k. λ)v ∗ (k.26) d k x x ηT b(k.29) In table 2 the behavior of particle states under the various transformations has been summarized.4 Bi-linear combinations In quantities such as currents and lagrangians often bilinear combinations of spinor fields are encountered. ˜ Top d(k. ˜ = 0|b(k)|A 8. λ) Top = ηT b(k. x (8.Discrete symmetries 68 Table 8. Therefore one has A|k = A|b† (k)|0 = A|T † T b† (k) T † T |0 ∗ = At |b† (k)|0 ∗ ˜ t = k|At . λ) e−ik·x (2π)3 2Ek ˜ d3 k ˜ ˜ ˜ ˜ ηT b(k.28) (8. λ) e−ik·x (2π)3 2Ek d3 k ˜ ˜ ηT b(k. λ |¯. λ).2: The transformation properties of physical states for particles (a) and antiparticles (¯). The result is = = λ (8. λ)u∗ (k.23) In the standard representation iγ5 C = σ2 and it is straightforward to apply this to the explicit field operator ψ(x) using i γ5 C u(k. p. −p. λ)v ∗ (k.33) (8. ˜ v ∗ (k. λ) e−ik·˜ (2π)3 2Ek d3 k ˜ ˜ ˜ ˜ ηT b(k. a state |a. λ) ψ t (x) −1 Top ψ(x) Top = i ηT γ5 Cψ(−˜) x = = ˜ u∗ (k. there are 16 independent of these bilinear combinations. λ)v ∗ (k.27) 3 = λ = λ = λ From this one obtains −1 ˜ Top b(k. λ) i γ5 C v(k. λ) e−ik·x (2π)3 2Ek (8. −p. −p. λ) eik·x + d† (k.30) (8. (8. λ). Note that applying an anti-unitary transformation such as Top one must take for the matrix element ˜ the complex conjugate. λ).

39) where Γµ (p′ . s . p) can be built from any combination of Dirac matrices (1.3: The behavior of the independent bilinear combinations of fermi fields under P. s′ |V µ (x)|p. of which the expectation value between momentum eigenstates can be simply found.37) (8. γ5 γ µ or γ5 ). C. 2M (8. ¯ In general the expectation value between momentum states can be more complicated.g. called form factors. Applying the results from the previous sections it is straightforward to determine the behavior of the combinations under P.40) where the coefficients are functions of invariants constructed out of the momenta. γ ν ].36) (8. p′ |V µ (0)|p = u(p′ )γ µ u(p). In many applications the expectation values of currents are needed. γ µ ...35) The x-dependence can be accounted for straightforwardly using translation invariance.5 Form factors Currents play an important role in field theory. for the vector current V µ (x).Discrete symmetries 69 Table 8. F. Note that the lagrangian density L (x) → L (−x) under Θ. in this case only q 2 = (p′ − p)2 . (q 2 ). C. 8. σ µν . V µ (x) = ei Pop ·x V µ (0)e−i Pop ·x . 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 µ ¯ photon field A (x) behaves in the same way as the vector combination ψ(x)γ µ ψ(x).41) 2M 2M . (8.38) (8. ¯ ′ ′ (8. F. 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 ). p′ . momenta (pµ or p′µ ) or constant tensors (ǫµνρσ or gµν ). (8. As an example consider the vector current for a point fermion. p) = γµ F1 (q 2 ) + iσµν q ν F2 (q 2 ). p′ |V µ (x)|p = u(p′ )Γµ (p′ . p)u(p) e−i (p−p )·x . and T. The 16 combinations of Dirac matrices appearing above are linearly independent. (q 2 ). This implies p′ |V µ (x)|p = e−i (p−p )·x p′ |V µ (0)|p . For instance for nucleons one has Γµ (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 . as well as under the combined operation Θ = PCT (see Table 3). e. ¯ V µ (x) =: ψ(x)γ µ ψ(x) :.

or relations based on hermiticity of the operator or P . p) that Γµ (p′ .43) = ei q·x u(p′ )γ0 Γµ (˜′ . • Parity p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . ¯ 2M (8. p) that Γµ (p′ .42) where q = p′ − p.Discrete symmetries 70 but that are eliminated because of relations between Dirac matrices.44) = p′ |Vµ (−˜)|˜ ˜ x p ∗ ˜x = ei q ·˜ [¯(˜′ )]∗ Γ∗ (˜′ . µ p ˜ (8. p′ )u(p′ )) u = ∗ † u† (p)γ0 Γµ (p. p)u(˜) ˜ x p ¯ p p ˜ p (8.1 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 ) + ′ . p)(iγ5 C)u(p). ¯ µ p ˜ Therefore time reversal invariance implies for the structure of Γµ (p′ . or relations that follow from the equations of motion. p)u(p) ¯ = p|V µ (0)|p′ ∗ = (¯(p)Γµ (p. p)(i γ5 C). C and T invariance. p)u(p) ¯ † † = p′ |Top Top V µ (x)Top Top |p ∗ (8. ¯ p ˜ Therefore parity invariance implies for the structure of Γµ (p′ . p). p)γ0 u(p). Therefore hermiticity implies for the structure of Γµ (p′ . p)u∗ (˜) up p µ p ˜ = ei q·x u(p′ ) (iγ5 C)Γ∗ (˜′ . p′ )u(p′ ) = u† (p′ )Γµ† (p. p)u(p) ¯ † † = p′ |Pop Pop V µ (x)Pop Pop |p ˜x = p′ |Vµ (˜)|˜ = ei q ·˜ u(˜′ )Γµ (˜′ . • Hermiticity p′ |V µ (0)|p = p′ |V µ† (0)|p = u(p′ )Γµ (p′ . p)γ0 . γ5 σµν = (1/2)ǫµνρσ γ ρ γ σ . p) = (i γ5 C)Γ∗ (˜′ .g. p′ )γ0 = Γµ (p′ . e. p ˜ • Time reversal p′ |V µ (x)|p = ei q·x u(p′ )Γµ (p′ . p) that γ0 Γµ† (p. p′ )γ0 u(p). e. the Gordon decomposition u(p′)γ µ u(p) = ¯ 1 u(p′ ) [(p′ + p)µ + iσ µν qν ] u(p). p) = γ0 Γµ (˜′ .45) Exercises Exercise 8.g.

0. p = (E. it would correspond to an electric dipole moment d = −e F3 (0)/2M (Interaction term d σ · E). (c) Show that if a term of this form would exist. |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 . (b) Show that in the interaction with the electromagnetic field. 0. 0. |q|). 4M 2 = F1 + F2 . −|q|/2). GE GM = F1 − Q2 F2 . 0. (b) Show that such a term is also not allowed by time-reversal symmetry.Discrete symmetries 71 (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 . L int = ejµ Aµ the charge is given by e GE (0) and the magnetic moment by e GM (0)/2M . σµν q ν F3 (q 2 ) 2M .2 (a) Calculate the current matrix element (using the explicit fermion spinors) in the Breit-frame in which q = (0. 0. 0. (c) Show that hermiticity of the current requires that F1 en F2 are real. Exercise 8. p′ = (E. Exercise 8.3 (a) Show that because of parity conservation no term of the form γ5 can appear in the current matrix element.

|ψS (t) = U (t. H]. (9. t0 ) ∂t (9. in which the states are time-independent. i ∂ |ψS ∂t ∂ i AS ∂t = H |ψS . t0 ) = H U (t. (9.5) (9. AH (t) = U −1 (t. |ψH (t) = |ψH . and the operators are time-dependent. (9.1 Time evolution as path integral U (t. i ∂ |ψH ∂t ∂ i AH ∂t ≡ = 0. t0 )|ψS (t0 ) . t QH (t)|q. AS (t) = AS and the states o are time dependent. [AH . t0 ). in which the operators are time-independent. (9.3) (9. ≡ 0. t′ . |q ′ .6) Of these the Heisenberg picture is most appropriate for quantum field theory since the field operators do depend on the position and one would like to have position and time on the same footing.4) ∂ U (t. t ≡ q|q. t0 ) AH (t0 ) U (t.Chapter 9 Path integrals and quantum mechanics 9. Consider the two (time-independent) Heisenberg states: |q.7) 72 . t . t′ ≡ q ′ |q ′ .2) (ii) Heisenberg picture. t′ QH (t′ )|q ′ . t0 ) = e−i(t−t0 )H or i Two situations can be distinguished: (i) Schr¨dinger picture.1) The time evolution from t0 → t of a quantum mechanical system is generated by the Hamiltonian.

Path integrals and quantum mechanics

73

and Schr¨dinger states o |q |q ′ QS |q ≡ q|q , QS |q ′ ≡ q ′ |q ′ .

(9.8)

Choose t as the starting point with |q = |q, t and QH (t) = QS and study the evolution of the system by calculating the quantum mechanical overlap amplitude q ′ , t′ |q, t = q ′ |e−i H(t −t) |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 ′ , t′ |q, t = = q ′ |e−i n H∆τ |q = q ′ | e−i H∆τ dq1 ...
n

|q (9.10)

dqn−1 q ′ |e−i H∆τ |qn−1 qn−1 | . . . |q1 q1 |e−i H∆τ |q .

The purpose of this is to calculate the evolution for an infinitesimal time interval. The hamiltonian is an operator H = H(P, Q) expressed in terms of the operators P and Q. 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|, 2π (9.11) (9.12)

where the transformation between coordinate and momentum space involves q|p = ei p.q . (9.13)

At least for a simple hamiltonian such as consisting of a kinetic energy term and a local potential, H(P, 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 , with the correction1 being of order (∆τ )2 . Then qj+1 , tj+1 |qj , tj ≈ = qj+1 |e−i K∆τ e−i V ∆τ |qj dpj qj+1 |pj pj |e−i K∆τ e−i V ∆τ |qj . 2π

(9.15)

By letting the kinetic and potential parts act to left and right respectively one can express the expectation values in terms of integrals containing H(pj , qj ), which is a hamiltonian function in which the operators are replaced by real-numbered variables. Combining the two terms gives qj+1 , tj+1 |qj , tj = = =
1 For

dpj i pj (qj+1 −qj ) −i H(pj ,qj )∆τ e e 2π dpj i pj (qj+1 −qj )−i ∆τ H(pj ,qj ) e 2π dpj exp (i ∆τ [pj qj − H(pj , qj )]) , ˙ 2π
1 1 1 [A, B] + [A, [A, B]] + [[A, B], B] + . . . 2 12 12

(9.16)

this, use the Campbell-Baker-Hausdorff formula, eA eB = eC with C = A+B+ (9.14)

Path integrals and quantum mechanics

74

or for the full interval q ′ , t′ |q, t =
τ

=

p Dq D exp i 2π

 dq(τ ) dp(τ ) exp i 2π
t′ t

t′ t

∆τ [pq − H(p, q)] ˙ (9.17)

dτ [pq − H(p, q)] , ˙

where Dq and Dp indicate functional integrals. 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. Before proceeding we also give the straightforward extension to more than one degree of freedom,
′ ′ q1 , . . . , qN , t′ |q1 , . . . , qN , t N

(9.18)
t′

=
n=1

Dqn D

pn 2π

exp i
t


n

pn qn − H(p1 , q1 , . . . , pN , qN ) ˙

.

9.2

Functional integrals

In this section I want to give a fairly heuristic discussion of functional integrals. What one is after is the meaning of Dα F [α], (9.19)

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, i.e. F [α] ∈ R. Examples are F1 [α] = α ≡ F2 [α] = α2 ≡ dx α(x), dx α2 (x), dx dy α(x) K(x, y) β(y), ≡ exp − 1 2 dx α2 (x) , dx dy α∗ (x) K(x, y) α(y) .

F3 [α, β] = α K β ≡ 1 F4 [α] = exp − α2 2

F5 [α, α∗ ] = exp (−α∗ K α) ≡ exp −

The kernels K(x, y) in the above examples can be (hermitean) operators acting on the functions, including differential operators, etc. Two working approaches for functional integration are the following: (i) the heuristic division of the space on which α acts into cells, i.e. α(x) → αx and F [α] = F (αx ) changes into a multivariable function, while Dα F [α] =
x

dαx N

F (αx )

(9.20)

becomes a multidimensional integral.

Path integrals and quantum mechanics

75

(ii) Write α in terms of a sum of orthonormal basis functions, α(x) = n αn fn and K(x, y) = ∗ m,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 )

(9.21)

again a multidimensional integral. In both cases N is an appropriately choosen normalization constant in such a way that the integral is finite. Note that procedure (i) is an example of the more general procedure under (ii). Consider the gaussian functional as an example. Using method 1 one writes 1 Dα exp − α2 2 =
x

dαx N

exp −

1 2

∆x α2 x
x

=
x

1 dαx exp − ∆x α2 x N 2 1 N 2π ∆x ≡ 1. (9.22)

=
x

The last equality is obtained by defining the right measure (normalization N ) in the integration. Physical answers will usually come out as the ratio of two functional integrals and are thus independent of the chosen measure. Using the second method and expanding in a basis set of functions one obtains for the (real) gaussian functional 1 Dα exp − α2 2 =
n

dαn N

exp −

1 αm αn 2 m,n

dx fm (x)fn (x)

=
n

=
n

dαn 1 exp − α2 N 2 n √ 2π ≡ 1. N

(9.23)

Having defined the Gaussian integral, the following integrals can be derived for a real symmetric or complex hermitean kernel K, 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) (9.24) (9.25)

(see Exercises). A useful property of functional integration is the translation invariance, Dα F [α] = Dα F [α + β]. (9.26)

As an important application of translation invariance we mention the identities 1 Dα exp − α2 − ωα 2 = exp 1 2 ω , 2 (9.27) (9.28)

Dα Dα∗ exp (−α∗ α − α∗ ω − ω ∗ α) = exp (ω ∗ ω) ,

Path integrals and quantum mechanics

76

Functional differentiation is defined as δ , α(y) = δ(x − y), δα(x) or in discretized form δ 1 ∂ ∂ with , αy = δxy , −→ δα(x) ∆x ∂αx ∂αx ∂ ∂ δ ∗ with , αn = δmn . −→ fn (x) δα(x) ∂αn ∂αm n Note that δ(x − y) =
n ∗ fn (x) fn (y). Examples of functional differentiation are

(9.29)

(9.30) (9.31)

δ δ α = dy α(y) = dy δ(x − y) = 1, δα(x) δα(x) δ αKβ = dy K(x, y) β(y) δα(x) 1 1 δ exp − α2 = −α(x) exp − α2 , δα(x) 2 2 δ exp (−α β) = −β(x) exp (−α β) . δα(x)

(9.32) (9.33) (9.34) (9.35)

For applications to fermion fields, we need to consider Grassmann-valued functions involving anticommuting Grassmann variables, i.e. θη = −ηθ, θ2 = 0. In principle the definitions of functionals is the same, e.g. the Gaussian-type functionals, F [θ, θ∗ ] = exp (−θ∗ θ) ≡ exp − F [θ, θ∗ ] = exp (−θ∗ K θ) ≡ exp − Integration for Grassmann variables is defined as dθ 1 = 0 and dθ θ = 1.
n θn fn (x)

dx θ∗ (x) θ(x) , dx dy θ∗ (x) K(x, y) θ(y) .

(9.36) where the coeffi-

This gives for the Gaussian functional integral after expanding θ(x) = cients θn are Grassmann variables the result Dθ∗ Dθ exp (−θ∗ θ) =
n ∗ dθn dθn

exp −

∗ θn θn n

=
n

∗ ∗ dθn dθn exp (−θn θn ) ∗ ∗ dθn dθn (1 − θn θn ) = 1.

=
n

(9.37)

We note that from the first to the second line one needs to realize that a pair of different Grassmann variables behaves as ordinary complex variables. In the expansion of the exponential (from second to third line), however, products of the same pair appear, which vanish. It is now easy to check that Dθ∗ Dθ exp (−θ∗ K θ) = det K. (9.38)

Path integrals and quantum mechanics

77

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 . ∗ (x) δη

= 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 ! Z Z t′ ′ ′ q , t |q, t = Dq exp i dτ Leff (q, q) , ˙ (9.45)
t

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) − ˙ ˙

i 2

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
′ ′

= =

p Dq D 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) . ˙

t = Dq q(ti1 ) . .57) dτ HI (τ ) U (τ. The interaction picture is defined as UI (t′ . t) = HI U (t′ . t) = one has U (0) (t′ . t). . t) = = ′ 1−i (−i) t t′ t dτ HI (τ ) U (0) (τ. tn }. . . The evolution of the (interaction) states is described only by HI . but also expectation of operators. at least if they appear time-ordered. . t) = 1 or U (t . ψI (t) ≡ ei H0 t ψS (t) = ei H0 t e−i Ht ψS (0) = e−iHI t ψS (0). if HI = 0 it is the Heisenberg picture and the evolution q ′ .53) i.Path integrals and quantum mechanics 78 Defining the time ordered product of operators T Q(t1 ) . The fast time evolution is described in H0 while HI is considered as a perturbation.54) (9. One works in the socalled interaction picture. q) . . . t) = U (1) ∞ n=0 ′ t′ = HI (t) ψI (t). Q(tn ) ≡ Q(ti1 ) .e. H0 ]. (9. Q(tn )|q. . ∂ ψI ∂t ∂ i AI ∂t i and ψI (t′ ) = U (t′ . AI (t) ≡ ei H0 t AS (t)e−i H0 t = ei H0 t AS e−i H0 t . t) ψI (t). Q(tin ). . q(tin ) exp i t t′ (9.4 An application: time-dependent perturbation theory From quantum mechanics one should be familiar with the procedure of time-dependent perturbation theory. t) ∂t with U (t.51) (9.50) Thus not only the quantum mechanical overlap of states can be calculated via a classical path integral. t) dτ HI (τ ) .e. 9. t (9. ∂ i U (t′ . i. t is described through the operators by H0 . one has q ′ .55) (9. = [AI . in which a separation is made of the hamiltonian H = H0 + HI . t) = 1 − i which can be solved by iteration. the lowest order leading to Fermi’s golden rule. .52) (9. t) − U (0) (t′ . (9. t) ≡ e−i H0 (t −t) . . t′ |T Q(t1 ) .56) (9. ′ (9.58) U (n) (t′ . t). where ti1 ≥ · · · ≥ tin is a permutation of {t1 . ˙ (9. . t′ |q.59) 1 t′ (t .49) dτ L(q. writing U (t′ .

t) = (−i) t dτ HI (τ ) U (n−1) (τ. t) − U (0) (t′ . . t) ′ (−i)n = n! n=0 ≡ T exp −i ∞ t′ t′ t′ dτ1 t t′ t dτ2 . . . t) t′ τ1 τn−1 = (−i) = n t dτ1 t t′ t′ dτ2 . . t t′ dτn HI (τ1 ) . . (9. . t) t′ U (n) (t′ . t) dτ HI (τ ) U (1) (τ. . . = 1 2 Z t′ dτ1 t Z Z t′ t t′ dτ2 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ) dτ2 Z t′ t + 1 2 t dτ1 HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ). HI (τn ) dτn T HI (τ1 ) . which can be rewritten (by interchanging in the second term the names of the integration variables) 1 = 2 Z t′ dτ1 t Z t′ t dτ2 [HI (τ1 ) HI (τ2 )θ(τ1 − τ2 ) + HI (τ2 ) HI (τ1 )θ(τ2 − τ1 )] . t . . = 1 2 Z t′ dτ1 t Z Z τ1 dτ2 HI (τ1 ) HI (τ2 ) t t′ + 1 2 dτ2 t Z t′ dτ1 HI (τ1 ) HI (τ2 ). the desired result. t dτn T HI (τ1 ) .Path integrals and quantum mechanics t′ t t′ t 79 U (2) (t′ . . . t) = 1 − i = (−i) . The integration Z t′ dτ1 t Z τ1 dτ2 HI (τ1 ) HI (τ2 ) t can also be performed by first integrating over τ2 but changing the integration limits (check!). HI (τn ) (9. dτ HI (τ ) U (0) (τ. HI (τn ). The time evolution operator. τ2 Now the integration can be extended by adding theta functions. t) + U (1) (τ. t) + U (1) (t′ . . . .61) dτ HI (τ ) . can be written as U (t . therefore. t (−i)n n! dτ1 t t dτ2 .60) The last equality is illustrated for the second term U (2) in the following. Thus we can write the integration as the sum of the two expressions (multiplying with 1/2).

64) (9. T = n ′ ′ J t′ = = Dq exp i t dτ [L(q. t q.62) t dτ [pq − H(p. one has q . T . Considering the source to be present between times t and t′ . a set |n of physical eigenstates. The Heisenberg state |q. ˙ J (9. q) + J(t) · q.66) . n (9. for g. q)] ˙ (9.s. i. i. T with q.65) J dq ′ dq Q′ . q)] . t is related to the Schr¨dinger state |q by |q. t = ei Ht |q . t′ q ′ . t t′ V =0 p exp −i Dq D 2π Dq D p exp i 2π dτ HI (q) exp i t t′ t dτ [pq − H0 (p. t′ |T exp −i t′ t′ t dτ HI (Q(τ )) |q. q) + J(τ )q] . t Q′ . t|n = q|n e−iEn t with q|n the time-independent wave function.Path integrals and quantum mechanics source experiment detector 80 T L t L+Jq t’ L T’ Figure 9.5 The generating functional for time ordered products By introducing a source-term. t′ |q. furthermore. T ′ |q ′ . T ′ |Q. Consider. 9. say. t = = V = q ′ . harmonic oscillator. 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. T = n q. t|n n|Q.e. Before and after these processes there is only the vacuum or ground-state |0 . We can again rewrite the time-ordered products as functional integrals as derived in the previous section. which in turn are embedded between an early time T and a future time T ′ . φn (q. t|Q. ˙ The combination of exponentials in the last step is allowed because we are simply dealing with classical quantities (not operators!). it is possible to switch on an interaction. q) → L(q. L(q. T < t < t′ < T ′ . making the electron) and the absorption of an electron (think of a detector). t|Q. for the physical states o q. t) φ∗ (Q) e+i En T . t |q. t′ |q. q ′ . ˙ ˙ physically pictured as.e.1: Physical picture of vacuum to vacuum amplitude in presence of a source term This is not a surprising result.63) for plane waves.

J (9.71) 9. Better is the use ˜ of imaginary time t = −it. such that t → −i∞ ′ t → i∞ ↔ ↔ ˜ t → −∞. .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 ′ ).Path integrals and quantum mechanics t 81 T t t’ T’ Figure 9. . T′ Z[J] = lim T → −i∞ T → i∞ ′ Dq exp i T dτ L q. T ′ |Q. . t′ ) q ′ . dq dτ + J(τ )q(τ ) .6 Euclidean formulation The above expression (Eq.2: Analytic continuation of boundaries T and T ′ We can project out the groundstate by an analytic continuation in the time.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 ) . t). t|Q. t′ |q. . J=0 (9. The generating functional precisely represents the vacuum to vacuum amplitude from initial (’in’) to final(’out’) situation in the presence of a source. δJ(tn ) hence the name generating functional. ˜ t′ → ∞. Since we have (neglecting multiplicative factors).67) = = φ0 (q. . 9. φ∗ (Q) 0 We define the generating functional Z[J] as Z[J] = Q′ . T = φ0 (q.68) (9. T → i ∞ and T ′ → −i ∞. The importance of Z[J] is that the time ordered product of operators can be obtained from it.70) one immediately finds δ n Z[J] δJ(t1 ) . in which case eiEn T → e−En ·∞ and the term eiE0 T is dominant. t 0 0out |0in J . (9. Q(tn )|0 . t) (9. thus T →i∞ lim eiE0 T q.

The differentiations in Z[J] and ZE [J] are related. In all of these cases the determinant of the (hermitean) operator K is defined as det K ≡ p Kp . τ dq ˜ dt ≡ −L q.74) − V (q) 2 (9. ensuring convergence for the functional integral ZE [J]. when L q.73) d˜ LE q. i τ −∞ ∞ −∞ dq d˜ τ + J(˜)q(˜) τ τ dq d˜ τ − J(˜)q(˜) τ τ . . for Grassmann valued functions.76) which is a positive definite quantity. . i dq . As discussed in the previous section the interesting quantities are obtained from functional differentiation with respect to the sources. Exercises Exercise 9. Dα Dα∗ exp (−α∗ K α) = for complex-valued functions and Dθ∗ Dθ exp (−θ∗ K θ) = det K. ˜ ˜ ZE [J] δJ(t1 ) .and Grassmann-valued functions. .1 Convince yourself by choosing an appropriate (orthonormal) set of functions that 1 Dα exp − α K α 2 for real-valued functions. det K Exercise 9. dq ˜ dt = 1 2 dq ˜ dt dq ˜ dt 2 − V (q). complex. ZE [J] = = where LE q. dq dt = 1 2 1 2 dq dt 2 Dq exp Dq exp − ∞ d˜ L q. 1 δ n Z[J] Z[J] δJ(t1 ) .Path integrals and quantum mechanics 82 In terms of the imaginary time we can write the Euclidean generating functional. + V (q).g.2 Check the examples of functional derivation for real-. .77) where the expressions have been divided by Z[J] and ZE [J] in order to get rid of dependence on multiplicative factors. ˜ dt (9. .n fm (x) Kmn fn (y). δJ(tn ) J = 0 ˜ t = it (9. y) = ∗ m. δJ(tn ) = (i)n J=0 1 δ n ZE [J] .75) = − LE q. 1 = √ .72) (9. (9. e. det K 1 . where Kp are the eigenvalues of the matrix Kmn in K(x. (9.

y) K −1 (y.80) Dα∗ Dα exp (−α∗ K α − α∗ ω − ω ∗ α) = 1 exp ω ∗ K −1 ω . z) = δ(x − z).79) (9.3 Prove the following relations 1 Dα exp − α K α − αω 2 1 = √ exp det K 1 ω K −1 ω . 2 (9.Path integrals and quantum mechanics 83 Exercise 9. det K Dθ∗ Dθ exp (−θ∗ K θ − θ∗ η − η ∗ θ) = det K exp η ∗ K −1 η .78) (9. . where dy K(x.

. xn ).Chapter 10 Feynman diagrams for scattering amplitudes 10. . .2) = ∞ −∞ d4 xE [LE (φ. . Π) (10. φ(x. ∂µ φ) + Jφ] (10. t) is considered as a set of quantum operators φx (t) in the Heisenberg picture and t′ φ . δJ(xn ) 84 ≡ G(n) (x1 . x .3) The Euclidean formulation is as before. . .1 Generating functionals for free scalar fields The generating functional for quantum fields is a generalization of the results in the previous section to a system with more degrees of freedom. 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 ) . φ(xn )|0 = (−i)n δ n Z[J] δJ(x1 ) .1) Dφ exp i σ d4 x L (x) and T′ Z[J] = lim T → i∞ T ′ → −i∞ Dφ exp − Dφ exp i T d4 x [L (φ. x. i. (10. ∂µ φ) − Jφ] (10. .4) J=0 . t ′ ′ ′ = = Dφ DΠ exp i t σ′ ˙ dτ d3 x Π(x)φ(x) − H (φ.e. t |φ. .

J(xn ).5) Z0 [J] for the free scalar field For the (free) scalar field lagrangian L = ← 1 1 ∂µ φ∂ µ φ − M 2 φ2 . which can be written as ∆(x) = ˜ with (k 2 − M 2 )∆(k) = 1 or ˜ ∆(k) = k2 1 1 1 = 0 2 . .11) Depending on the path choosen in the complex k 0 plane (see exercise 7. .7) where we made the choice Z0 [0] = 1 and ∂µ ∂ µ + M 2 ∆F (x − y) = −δ 4 (x − y). with Z0 [0] = 1.13) . . . (10. . . . .10) J=0 In order to determine ∆F . ∆F (the socalled Feynman propagator) is the Green’s function of the Klein-Gordon equation. = 0 2 2 2 2 −M (k ) − E 2 (k ) − k − M (10. As discussed in the previous section (exercise 9. .3) one finds 1 Z0 [J] = exp − i 2 d4 x d4 y J(x) ∆F (x − y) J(y) ≡ exp (i W0 [J]) . (2π)4 (k 0 + iǫ)2 − E 2 (10. xn ) = x2 xn . .Feynman diagrams for scattering amplitudes 85 for which we introduce the picture x1 G(n) (x1 .e. These Green functions appear in the expansion Z[J] = n in n! d4 x1 . Furthermore. (10. . one immediately sees that i∆F (x1 − x2 ) = δ 2 Z0 [J] (−i)2 Z0 [0] δJ(x1 ) δJ(x2 ) = G0 (x1 − x2 ) ≡ (2) x1 x2 (10.4) one distinguishes the retarded Green’s function ∆R (x) = d4 k e−i k·x . consider the general solution of ∂µ ∂ µ + M 2 ∆(x) = −δ 4 (x).12) d4 k −i k·x ˜ e ∆(k) (2π)4 (10.8)  1 d4 x  φ (∂µ ∂ µ + M 2 ) φ − Jφ 2 K  (10. 2 2 → (10.9) i. . xn ) J(x1 ) .6) the generating functional is (using ∂µ = − ∂µ ) given by   Z0 [J] =  Dφ exp −i where Kxy = (∂µ ∂ µ + M 2 ) δ 4 (x − y). (10. d4 xn G(n) (x1 .

the second is to explicitly consider 0|T φ(x1 ) φ(x2 )|0 . (10.19) φ ∂E ∂E − M 2 φ + Jφ 2 1 d4 xE d4 yE J(x) (−i ∆F (x − y)) J(y) .23) k0 =i∞ k0 =−i∞ d4 k e−i k·x . (10. ∆(x) = − d4 k e−i k·x .18) Thus we see various solutions.16) C where C is a closed contour in the complex k 0 -plane. t) = 0 for t > 0.22) (10.4).e. t) = 0 for t < 0. φ+ (0)] = (2π)3 2E are also shown in exercise (7. . The first possibility is to consider the well-defined Euclidean formulation.21) ∆F (x) = = −i 0 kE =−∞ d4 kE e−i kE ·xE 2 (2π)4 kE + M 2 (10. (2π)4 k 2 − M 2 (10. The contours for i ∆(x) i ∆+ (x) i ∆− (x) = [φ(x). (2π)4 k 2 − M 2 (10. 10.20) = exp − 2 where or (see fig. the advanced Green’s function ∆A (x) = satisfying ∆A (x. the causal Green’s function ∆C (x) = and finally the Green’s function d4 k e−i k·x = (2π)4 (k 0 )2 − E 2 + iǫ ∆(x) = PV d4 k e−i k·x . φ(0)] = i∆+ (x) + i∆− (x) d3 k e−i k·x = [φ+ (x). 4 (k 0 − iǫ)2 − E 2 (2π) (10. i. 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).15) e−i k·x d4 k . The solution of the homogeneous equation ∂µ ∂ µ + M 2 ∆(x) = 0 can also be written as an integral in k-space.17) d4 k e−i k·x . in the (well-defined) Euclidean formulation starting with LE = 1 (∂E φ)2 + 2 M 2 φ2 2 µ 2 2 Jφ with the equation of motion (∂E ) − M φ(x) = −J(x). ∆F = ∆C . φ− (0)] = (2π)3 2E d3 k e−i k·x = [φ− (x).14) They are solutions of the inhomogeneous equation.1 for countours) µ (∂E )2 − M 2 (i ∆F (x − y)) = −δ 4 (x − y) 0 kE =∞ (10. the generating functional can be written 1 µ µ Z0 [J] = Dφ exp d4 xE (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. (2π)4 k 2 − M 2 where the latter contour can be deformed to the contour for ∆C . µ 1 • Firstly.Feynman diagrams for scattering amplitudes 86 satisfying ∆R (x. (2π)4 k 2 − M 2 + iǫ (10.

24) Knowing the explicit form of Z0 [J] it is straightforward to calculate the 4-points Green’s function. because this is the only quantity entering Z0 [J].1: Contour in the k 0 plane for the Feynman propagator in Euclidean and Minkowski space • As the second possibility. Neglecting multiplicative factors or equivalently assuming that Z0 [0] = 1. x3 . x4 ) = 0|T φ(x1 ) φ(x2 ) φ(x3 ) φ(x4 )|0 1 δ4 exp − i J ∆F J = (−i)4 δJ(x1 ) . . x2 . δJ(x4 ) 2 J=0 = − [∆F (x1 − x2 ) ∆F (x3 − x4 ) + ∆F (x1 − x3 ) ∆F (x2 − x4 ) + ∆F (x1 − x4 ) ∆F (x2 − x3 )] . we have G0 (x1 . φ+ (y)]|0 = θ(x0 − y 0 ) i ∆+ (x − y) − θ(y 0 − x0 ) i ∆− (x − y) (10. We find (see also section 7.3) i ∆F (x − y) = 0|T φ(x) φ(y)|0 = θ(x0 − y 0 ) 0|φ(x) φ(y)|0 + θ(y 0 − x0 ) 0|φ(y) φ(x)|0 = θ(x0 − y 0 ) 0|φ+ (x) φ− (y)|0 + θ(y 0 − x0 ) 0|φ+ (y) φ− (x)|0 = i ∆C (x − y). . = θ(x0 − y 0 ) 0|[φ+ (x). φ− (y)]|0 − θ(y 0 − x0 ) 0|[φ− (x). or diagrammatically (4) (10. (4) (2) It should be clear that G0 can be expressed in terms of G0 = i ∆F . the same result as above.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 0 0 1 1 1 0 1 0 1 0 1 0 1 0 87 Figure 10. we use the fact that i∆F (x − y) = G(2) (x − y) = 0|T φ(x)φ(y)|0 to calculate ∆F via the time-ordered product.25) x1 x2 x3 = x4 + + .

with propagators (i ∆F ) being connected to vertices (i LI ) according to the above expression for the functional Z[J].34) Introducing a vertex point to which four propagators are connected. the generating functional can be written. x3 . (10.26) Dφ exp i = exp i d4 z LI 1 δ i δJ(z) d4 x d4 y J(x) ∆F (x − y) J(y) (10. Consider as an example the interaction LI (φ) = − g 4 φ 4! in the scalar field theory discussed sofar. x2 ) (4) (0) (10. x4 ) = To first order in g one has Z (g ) [J] = = 1 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 .33) G(4.g 0 ) = G0 = 1.g ) (x1 .g ) (x1 . x2 ) = 0 0 x1 + + x2 (10.31) = i ∆F (x1 − x2 ) = G(2. (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 . (10. Z[J] = = Dφ exp i exp i d4 z LI (10. ≡ −i g d4 z (10.30) and in zeroth order G(0. x2 .32) (10. x4 ) = G0 (x1 . x3 .27) When interactions are present. L (φ) = L0 (φ) + LI (φ). i.29) This expression will be the one from which Feynman rules will be derived.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.e.35) z . To zeroth order in the coupling one has 0 1 Z (g ) [J] = Z0 [J] = exp − i 2 J ∆F J . x2 .Feynman diagrams for scattering amplitudes 88 10. (2) G0 (x1 .

xn ) = (−i)n−1 c which are denoted δ n W [J] δJ(x1 ) . . . i. . (10.41) in terms of socalled connected Green’s functions.37) G(2. . . x2 ) = 1 z x1 x2 + 1 2 x1 z x2 . did not contain parts that could be written as products of simpler Green’s functions. (10. . G(n) (x1 . J(xn ). . In the free case we have seen that the exponent of Z[J] contained all essential information. . . . Diagrammatically this exponent only contained the two-point function.Feynman diagrams for scattering amplitudes 89 one has J Z (g1 ) [J] = 1 8 1 − 4 J J 1 + 24 J Z0 [J]. xn ) J(x1 ) . This remains also true for the interacting case. . .36) J J from which one obtains G(0. xn ) = x2 c xn Specific examples are G(0) = c G(1) (x1 ) = c G(2) (x1 .38) The result for the 4-points Green’s function is left as an exercise. xn ) J(x1 ) . .g ) (x1 . .40) in−1 n! d4 x1 . . δJ(xn ) . (10. Z[J] = n in n! d4 x1 . . J=0 (10. 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]. . d4 xn G(n) (x1 . which also was the only Green’s function for which the diagram was connected. . .42) x1 G(n) (x1 . . x2 ) = c (vacuum bubble) (tadpole) (connected propagator) . . 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]. . . .39) We have calculated the function up to first order in the coupling constant for φ4 -theroy. d4 xn G(n) (x1 .g 1 ) = 1 8 z 1 8 . J(xn ) c (10. . (10.e. . (10.

1 2! Z[0] = 1 + + + .. (iv) If δZ[J] = i 0|φ(x)|0 = 0.. + +i d4 x1        1 2! + 1 3! +i d4 x1 + 1 2! i2 2! d4 x1 d4 x2              + . These can be divided out. 2! 3! +i d4 x1 + i2 2! d4 x1 d4 x2 ... In the expansion of Z[J]/Z[0] one (by definition) has the socalled source (n) connected Green’s functions Gsc . 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 + + . δJ(x) J=0 . but is easily illustrated by writing down the first few terms. (1) (1) (iii) Gsc = Gc .. (ii) Note that Z[0] = exp(iW [0]) appears as a multiplicative factor..Feynman diagrams for scattering amplitudes 90 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).

out . In line with the consideration of the generating functional representing the vacuum to vacuum amplitude we consider fields φin . Sβα ≡ β. in|p′ . Proof: 0. out|p′ . t = −∞ φin (x) t = +∞ φout (x) φ(x) where φin and φout transform under the Poincar´ group as scalar fields and satisfy the homogeneous e Klein-Gordon equation with the physical mass M . x1 G(4. e. . we have W0 [J] = − 1 2 d4 x d4 y J(x) ∆F (x − y) J(y). out| = eiϕ0 0. (10. pn . Also for free field (Heisenberg) operators a distinction is made between φin and φout . in|SS † |α. ∂µ ∂ µ + M 2 φin (x) = 0. .44) J=0 For the interacting theory. (2) (10. x2 .g. x4 10. in = β. thus β. . (2) The one-particle state is invariant (conservation of energy and momentum. . in = |p1 . x3 . then (n) (n) Gsc = Gc for n ≤ 3. out|α. implying the absence of tadpoles.46) The properties of the S-matrix are (1) The vacuum is invariant or |S00 | = 1. in = p.3 Interactions and the S-matrix The S-matrix The S-matrix transforms initial state free particle states (in-states) |α. out|α. in|S = β. in = |α. in| (choose ϕ0 = 0). in|S|p′ .47) . Next. the vacuum expectation value of the field φ(x) is zero. Proof: α. . out = p|p′ .45) x2 as the only surviving diagram (see exercises). . φout and the interpolating field φ(x). out = |p′ . out| ⇔ |α.g ) (x1 . out = δαβ ↔ SS † = 1.43) 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 ). x4 ) = c 1 x3 z (10. in|S = α. (10. in|S = 0. one finds for the connected 4-point Green’s function at order g. in = p. in ⇔ β. in into final state particle states (out-states) |β.e. out|p′ . this will be translated to the action on fields. in = S|α. p. (10. translation invariance). For the free scalar theory. (3) S is unitary (it conserves the scalar product from initial to final state). p′ . in = p. out| and S † |α. out (suppressing except momenta all other 1 m quantum numbers).Feynman diagrams for scattering amplitudes 91 i. considered explicitly. out .

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. (2π)3 2E (10.51) (10. The relation between S and Z[J] To establish this relation.53) 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). The time evolution operator (see Eq.50) = φ(x) = √ Zφin (x) − √ Zφout (x) − d4 y d4 z ∆R (x − y) K(y.Feynman diagrams for scattering amplitudes 92 while φ satisfies the inhomogeneous Klein-Gordon equation with the bare mass M0 (this is the mass appearing in the lagrangian L0 ). 9. the source term Jφ in the lagrangian density is treated in the interaction picture. √ Although the above. in|Sφout Finally we check that as expected S does not spoil Poincar´ invariance.62) is then T UT [J] = T exp i ′ T′ T d4 x J(x)φ(x) . z) φ(z) d4 y d4 z ∆A (x − y) K(y.and out-fields is: φin (x) = S φout (x)S −1 .e. (10. as it stands. Proof: β. 2 ˜ ∂µ ∂ µ + M 2 φ(x) = J(x) + (M 2 − M0 )φ(x) = J(x) √ ˜ φ(x) = Zφin (x) − d4 y ∆R (x − y) J(y) (10. (10. φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . implies the strong (operator) convergence φ(x) → Zφin (x). e. this can actually not be used as it would imply [φ(x).54) . β. a) = S.g.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. i. φin (y)] = iZ ∆(x − y). a)SU −1 (Λ. which in a later stage (renormalization) will become more important. (10. The relation between S-matrix and in. Considering plane waves one sees from √ √ 0|φ(x)|p = lim Z 0|φout (x)|p = Z e−i p·x t→∞ √ √ = lim Z 0|φin (x)|p = Z e−i p·x t→−∞ ↔ that identical normalization of (single-particle) plane waves in initial and final state implies the same wave function normalization Z for in and out fields. The convergence therefore must be weakened to t→−∞ √ Z α|φf (t)β (10.52) where Z is a constant. a causality condition that can be proven to imply the absence of interactions. 2 ∂µ ∂ µ + M0 φ(x) = J(x). out|φout = β. S is invariant under e Poincar´ transformations: U (Λ.49) The field φ can be expressed in φin/out using retarded or advanced Green’s functions. z) φ(z). in|φin S → φin S = S φout (x). φ(y)] = Z [φin (x).

φ(y)] f (y) e e . S U [J]] = √ Z δ S U [J]. δU [J] δJ(x) = = √ i Z U [J] φin (x) − √ i Z φout (x) U [J] − d4 y d4 z ∆R (x − y) K(y. : =i R d4 y ∆(x − y) f (y) : e R d4 z φ(z) f (z) :.56) Since δU [J]/δJ(x) satisfies the same equations as φ(x) and we know the limits. B] is a c-number).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). with the use of the Baker-Campbell-Hausdorff formula e−B A e+B = A + [A. B]eB . we can. δJ(x) ∞ or for U [J] ≡ U−∞ [J] ′ (10. Thus applied to the field φ(x) = φ+ (x) + φ− (x).59) (10.60) e R d4 y φ+ (y) f (y) R d4 z φ− (z) f (z) R d4 z φ+ (z) f (z) d4 y [φ(x). z) d4 y d4 z ∆(x − y)K(y. one has [A.Feynman diagrams for scattering amplitudes 93 and satisfies the property T ′ δUT [J] t = i UtT [J] φ(x)UT [J]. z) δU [J] δJ(z) δU [J] d4 y d4 z ∆A (x − y) K(y. Noting that 0| : eA : |0 = 1 it follows that F [J] = 0|S U [J]|0 = 0|U [J]|0 = 0out |0in ∝ Z[J].e. Thus S U [J] =: exp 1 √ Z d4 x d4 y φin (x) K(x. φ(x).61) e R φ+ f where the normal ordered expression : e φ f : is used. y) δ δJ(y) : F [J]. z) . just as we did for φ(x). (10. z) δU [J] .63) .62) where F [J] is some (arbitrary) functional. δJ(z) (10. which is equal to the expression e in which the creation operators are placed left of annihilation operators. provided [A. : e R d4 z φ(z) f (z) R d4 x φ− (x) f (x) B C B C (10. R φ− f (10. δJ(z) Taking the difference between the two expressions √ Z (φout (x) U [J] − U [J] φin (x)) =i or d4 y d4 z (∆R (x − y) − ∆A (x − y)) K(y. (10. express it in terms of advanced and retarded Green’s functions. t i Ut∞ [J] φ(x)U−∞ [J] (10. B] (for the case that [A. δJ(z) (10. e e ] = [A. C] are c-numbers. φ(x). eB ] = [A.58) In order to find a solution to this equation note that. B] and [A. e = i. B + C]e e .57) i [φin (x). [A.

. using that φin (x) = d3 k ∗ a(k) fk (x) + a† (k) fk (x) . . i.) iKx ..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.. . n (2π)4 δ 4 (p1 + . They start with the propagator (i∆F (k)) in momentum space. xn ). . . . y) δ δJ(y) : Z[J] Z[0] . pn ) = i=1 d4 xi ei pi ·xi G(n) (x1 . |S|p. . = . iKx′ . . . d4 x . . c (n) (10. . . p′ . (10. . S =: exp 1 √ Z d4 x d4 y φin (x) K(x. . . . ¯ ¯  −i  (n) G (p1 .68) 10.67) and the (amputated) Green’s functions Γ(n) (p1 . . which is determined by the inverse of the operator found in the quadratic term in the lagrangian i. .65) we get the Lehman-Symanzik-Zimmerman (LSZ) reduction formula.Feynman diagrams for scattering amplitudes 94 while for J = 0 one has U [0] = 1..e. + pn ) G(n) (p1 . Sf i = δf i − i (2π)4 δ 4 (Pi − Pf ) Mf i . . . . . . . .δ. u(p).e. y). Usually we are interested in the part of the S-matrix describing the scattering. . . . . . x. J=0 (10. . . an S-matrix element between momentum eigenstates in initial and final states is found by considering those source-connected Green’s functions (action of δ/δJ on Z[J]/Z[0]) where the external sources J(xi ) are replaced by the particle wave functions (which are the result of acting with φin (xi ) on momentum eigenstates). Note that the Green’s function connecting the external point xi with the bubble is annihilated by K(x. . pn ) =   n j=1 where ∆(p) is the Fourier transform of the (full) propagator ∆F (x). . i.64) Therefore. u(p) and v (p) for fermions and ǫµ (p) for vector fields). . (λ) v(p). see exercise 10. iKx′ = (2x′ + M 2 ) precisely annihilating an external propagator i∆F in the Green’s function. fp′ (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. . which is obtained considering only connected diagrams.+ 1 √ Z ∗ d4 x′ . k = i k 2 − M 2 + iǫ . (2π)3 2E n (10. . .66) where Gc is the connected Green’s function. pn ). 1 for scalar case. . ∆(pj ) (10. Explicitly. .e. ←− − → × G(n) (x′ . .2). Next we introduce the Fourier transform (after extracting a momentum conserving delta function coming from translation invariance.. fp (x) .

vertices and external particle wave functions in that diagram. At these vertices each line can be assigned a momentum. (Rule 3) Carry out the integration over all internal momenta (keeping track of momentum conservation at all vertices!) (Rule 4) Add a symmetry factor 1/S corresponding to permutation of internal lines and vertices (keeping external lines fixed). or calculating full Green’s functions external lines are treated as propagators. If problems arise go back to the defining expression for the generating function in 10.27. = −i g (multiplied with 4! corresponding to the allowed number of permutations of identical particles). For the symmetry factor consider the examples (given in G. which is summarized in the following rules: (Rule 1) Start with external legs (incoming particles/outgoing particles) and draw all possible topologically different connected diagrams. to be precise iLI vertices in momentum space are introduced. Note that in calculating amputated Green’s functions external lines are neglected. Veltman. ’t Hooft and M. For the interaction terms in the lagrangian. Corresponding to external particles wave functions are introduced k k 1 1 In order to calculate the connected amplitude −i Mf i appearing in the S-matrix element these ingredients are combined using Eqs 10.69) LI (φ) = − φ3 − φ4 . = −i g .66 and 10. which cancels the factor 1/2 in the quadratic piece. but overall momentum conservation at a vertex is understood. the factor 2 corresponds to the two-points Green’s function having two identical ends).Feynman diagrams for scattering amplitudes 95 (in fact the inverse of the operator found in the quadratic term is for real scalar fields also multiplied by a factor 2.27. 3! 4! The vertices are: = −i f Consider first the lowest order self-energy diagram. 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. Diagrammar) in the case of the interaction terms g f (10.

i. Dividing by vertex factors and permutations of identical vertices. i δJ(z) i δJ ∗ (z) d4 x d4 y J ∗ (x) i∆F (x − y) J(y) . Finally divide by the number of permutation of the points that have identical vertices (here 2!).e.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. denoted k = k2 i − M 2 + iǫ Note that in this case the propagator does not have identical ends. or equivalently as independent fields φ and φ∗ .Feynman diagrams for scattering amplitudes 96 Draw two points corresponding to the two vertices and draw in each of these points the lines coming out of the vertices: Now count in how many ways the lines can be connected with the same topological result. after that line 2 in three ways. The total result is 6×3×2 1 1 = = . S 3! 3! 2! 2 As a second example consider the diagram 1 2 There are three vertices. but it connects a source with its complex conjugate and therefore is oriented. which have been included in the definition of vertices (here 3! for each vertex). Then there are two ways to connect the remaining lines such that the desired diagram results. J ∗ ] = Dφ Dφ∗ exp i d4 z LI d4 x φ∗ −∂µ ∂ µ − M 2 φ + LI (φ) + J ∗ φ + Jφ∗ δ 1 1 δ . the result is 1 6×4×6×3×2 1 = = . (10. External line 1 can be attached in six. Divide by the permutational factors of the vertices. there is no combinatorial factor like in the scalar case. 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. S 3! 3! 4! 2! 2 Complex scalar fields The case of complex scalar fields can be considered as two independent fields. .

. − 1 2 dx dy dx′ dy ′ η(x)S(x − y)η(y) η(x′ )S(x′ − y ′ )η(y ′ ).g. η). η] = Dψ Dψ exp i d4 z LI exp − d4 x ψ (i/ − M ) ψ + LI (ψ) + ηψ + ψη ∂ 1 δ 1 δ . but the anticommutating properties of Grassmann variables imply some additional minus sign in Feynman diagrams..Feynman diagrams for scattering amplitudes 97 Dirac fields For fermions the generating functional is given by Z[η.. which arises from the quadratic term in exp i d4 z LI . which is the solution of (i/ − M )SF (x) = −δ 4 (x) ∂ (i. The two-points Green’s function 0|T φ(x) φ(y)|0 contains a fermionic loop contribution. (Rule 6) Each closed fermion loop gets a sign −1.e. = + .. (10.73) = 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. 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 . e. (2π)4 p2 − M 2 + iǫ and the (oriented) propagator in Feynman diagrams involving fermions is i p j (10. in e− e− → e− e− scattering (Møller scattering) the lowest order contribution is - (see next section for e-e-γ vertex). The latter rule is illustrated in the example of an interaction term LI = : gψ(x)ψ(x)φ(x) : in an interacting theory with fermionic and scalar fields. − g2 2 dz dz ′ δ2 δ2 ′ ) δη(z ′ ) δη(z) δη(z) δη(z and the quadratic term in Z0 [η.71) = exp i where iSF is the Feynman propagator for fermions.− i δη(z) i δη(z) d4 x d4 y η(x) iSF (x − y) η(y) .72) (10. namely (Rule 5) Feynman diagrams which only differ by exchanging identical fermions in initial or final state have a relative minus sign.

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 gµν 1 − 1− k 2 + iǫ λ kµ kν . The wave functions for fermions are given by ∝ g 2 S(z − z ′ ) S(z ′ − z) = −g 2 iS(z − z ′ ) iS(z ′ − z).74) L = − Fµν F µν − (∂ · A)2 + ψ(i/ − M )ψ − e ψγ µ ψ Aµ . Vector fields and Quantum Electrodynamics As the most important example of vector fields consider the lagrangian density for quantum electrodynamics (QED). 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. 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. 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 . (k 2 + 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. 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. λ 1 ∂ (10. writing down the propagators till the beginning of the ¯ line (incoming fermion u or outgoing antifermion v) has been reached.Feynman diagrams for scattering amplitudes 98 The result is which contains an extra minus sign as compared to a bosonic loop. 1 k 2 − M 2 + iǫ λ (k 2 − M 2 + iǫ)(k 2 − λ M 2 + iǫ) leads to . 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 . Particular choices of λ are λ = 1 (Feynman propagator or Feynman gauge) and λ = ∞ (Landau gauge).

s3 )(−ie)γ µ u(k.Feynman diagrams for scattering amplitudes 99 10. s)γν 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. s′ ) ¯ u s. s2 )|2 ¯ q2 (10. q 4 µν [using that (¯(k ′ )γµ u(k))∗ = u(k)γµ u(k ′ )]. s3 )γ µ u(k.s4 ) and averaged over spins in the initial state (1/2 × 1/2 × s1 .79) . The diagram and momenta and the commonly used invariants (Mandelstam variables) for a 2 → 2 scattering process are 1 k 3 k’ s t u = (k + p)2 = m2 + M 2 + 2 k · p = (k ′ + p′ )2 = m2 + M 2 + 2 k ′ · p′ = (k − k ′ )2 = q 2 = 2 m2 − 2 k · k ′ = (p − p′ )2 = 2 M 2 − 2 p · p′ p 2 p’ 4 s = (k ′ − p)2 = m2 + M 2 − 2 k ′ · p + t+u= m2 = 2 m2 + 2 M 2 i = (k − p′ )2 = m2 + M 2 − 2 k · p′ The scattering amplitude is given by −iM = u(k ′ .s2 . s1 ) u e2 u(p′ . s2 ). To lowest order (∝ α = e2 /4π) only one diagram contributes.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. ¯ q2 (10. If we are interested in the scattering process of an unpolarized initial state and we are not interested in the spins in the final state we need |M |2 summed over spins in the final state ( s3 . s4 )γµ u(p. s′ )γµ u(k.s3 . s4 )(−ie)γ ν u(p. where u ¯ L(m) µν = 1 2 u(k ′ .77) Combining Lµν and Lµν (M) one obtains L(m) Lµν (M) = 2 s2 + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 µν and M 4π 2 (m) (10. t2 (10.78) = 2 α2 2 s + u2 + 4t(M 2 + m2 ) − 2(M 2 + m2 )2 .5 Some examples eµ scattering The first example is the electromagnetic scattering of an electron and a muon. s1 ) ¯ −igµν u(p′ .76) e (m) µν (M) L L . s)¯(k.s4 |¯(k ′ .s2 ) which can be written as (see also chapter 4) |M |2 = = 1 4s 4 1 .

s3 . 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. s3 )γ µ v(p′ .s4 = 1 e4 1 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 and = 2 α2 2 t + u2 + 4s(M 2 + m2 ) − 2(M 2 + m2 )2 . s3 )γ µ v(p′ . .s2 . s1 ) u(k. The diagram. s1 )|2 ¯ s = 1 2s × v (k ′ . momenta and invariants are e− e+ → µ− µ+ scattering k p s t = = (k + k ′ )2 = (p + p′ )2 (k − p)2 = (k ′ − p′ )2 (k − p′ )2 = (k ′ − p)2 k’ p’ u = The scattering amplitude squared (spins summed and averaged) is given by |M |2 = 1 4s e s2 4 1 . s2 ) ¯ ¯ 1 2s u(p. s4 )γ ν u(p. (c) Do the same for the connected Green function Gc . for instance Møller scattering (e− e− → e− e− ) and Bhabha scattering (e− e+ → e− e+ ) are related using crossing symmetry. the masses. Similarly.80) Note the similarity in the amplitudes for eµ scattering and e− e+ → µ− µ+ . s1 )γν v(k ′ . To lowest order (∝ α = e2 /4π) only one diagram contributes. s4 ) v (p′ . s3 ) ¯ ¯ 1 .s2 3 .Feynman diagrams for scattering amplitudes 100 The second example is the annihilation of an electron pair and creation of a muon pair. (b) Use the definition of the source-connected Green function Gsc and show that indeed sourceconnected diagrams survive.1 (a) Give in diagrammatic notation the full Green functions G(4) (x1 . s4 ) u e2 v (k ′ . (4) (4) . Exercises Exercise 10. Basically the same diagram is calculated and the result is the same after the interchange of s ↔ t. . s2 )γµ u(k. s2 )γµ u(k.s4 |¯(p. .2. This is known as crossing symmetry. s2 (10.

.37 and 10. .3 Show that the combinatorial factors found using the rules given in section 10. . . t and u. + pn ). pn ).2 (a) Show that the translation properties of the fields and the vacuum imply G(n) (x1 + a. . . . . xn ). xn + a) = G(n) (x1 . . .38. |M |2 = T11 + T22 − T12 − T21 . xn ) ≡ (2π)4 δ 4 (p1 + . xn ) ∝ (2π)4 δ 4 (p1 + . It is of the form −i M = A1 − A2 . . Exercise 10. . Exercise 10. . e(p1 ) + e(p2 ) → e(p′ ) + e(p′ ) in lowest order in α. 1 2 (b) Calculate the quadratic pieces and interference terms. Show that i the amplitude is symmetric under the interchange of t ↔ u. . . . + pn ) G(n) (p1 . hence we can write n i=1 d4 xi ei pi ·xi G(n) (x1 . . . .4 (a) Write down the Feynman diagrams contributing to electron-electron scattering. which means overall momentum conservation in Green functions in momentum space.3 reproduce for the diagrams precisely the combinatorial factors that appear in Eqs 10. Express the contributions in invariants s. . in the amplitude (Tij = A∗ Aj ). . . . . (b) Show (by using x1 as shift-varible) that this implies that n i=1 d4 xi ei pi ·xi G(n) (x1 . .Feynman diagrams for scattering amplitudes 101 Exercise 10.

−q). pb ) satisfying p2 = m2 and p2 = m2 . For two particles. pn ). . pb we start with the four vectors pa = (Ea .3) which is useful because the total 4-momentum of the final state usually is fixed by overall momentum conservation. To be specific let us consider two frequently used frames. (11. . . i. Its square root. It is determined by the energy-momentum four vector p = (E. In that case cm pa = (Ea .6) (11. . pn ) = (2π)4 δ 4 (P − pi ) dR(p1 . pa ) and pb = (Eb .1) = 3 2E (2π) (2π)4 (proven in Chapter 2).1 kinematics in scattering processes Phase space The 1-particle state is denoted |p . s = (p1 + . .5) √ is referred to as the invariant mass squared. (2π)3 2Ei i=1 (11. a a b b the quantity s = P 2 = (pa + pb )2 . . (11. 102 . (11. . . p1 . i n d3 pi . + pn )2 . Here s is the invariant mass of the n-particle system.Chapter 11 Scattering theory 11. . . This is generalized to the multi-particle phase space dR(p1 . for instance for determining the threshold energy for the production of a final state 1 + 2 + . the 1particle phase space is d3 p d4 p θ(p0 ) (2π)δ(p2 − m2 ). and the total momentum four-vector P = pa + pb . q). . .2) (11. (11. In the CM frame the threshold value for s obviously is n 2 sthreshold = i=1 mi . The phase space is determined by the weight factors assigned to each state in the summation or integration over states. The first is the CM system.4) For two particle states |pa . A physical state has positive energy.7) pb = cm (Eb .e. . s is for obvious reasons known as the center of mass (CM) energy. pn ) = and the reduced phase space element by dR(s. It is a useful quantity. . . p) which satisfies p2 = E 2 −p2 = m2 . . + n.

= 2 s λ(s.12) pb = (mb . 2mb λ(s.11) (11.9) (11. m2 .16) . 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. m2 ). for instance.8) (11. 0). m2 . m2 ) is a function symmetric in its three arguments.14) One can. 2 s s − m2 + m2 a b √ . (11. which in the specific case a b also can be expressed as λ(s. In that case trf pa = (Ea .13) (11. p2 ) = = where λ12 denotes λ(s. Prove that trf Ea = |ptrf | = a s − m2 − m2 a b .10) The function λ(s. m2 ) = 4(pa · pb )2 − 4p2 p2 . m2 . a (11.15) Explicit calculation of the reduced two-body phase space element gives dR(s. ptrf ). Also in this case one can express the energy and momentum in the invariants. a a b b The second frame considered explicitly is the socalled target rest frame in which one of the particles (called the target) is at rest. p1 . mb and s).Scattering theory 103 It is straightforward to prove that the unknowns in the particular system can be expressed in the invariants (ma . m2 . 2mb (11. m2 . m2 ) a b . Prove that |q| = cm Ea cm Eb (s − m2 − m2 )2 − 4 m2 m2 a a b b = 4s s + m2 − m2 a b √ = . 4s (11. 4π s 4π 8π s 4π (11. trf Ea (threshold) = 1 2 mb ( i mi )2 − m2 − m2 a b . 1 2 d(E1 + E2 ) √ |q| dΩ(ˆ) q δ( s − E1 − E2 ) 2 (2π) 4(E1 + E2 ) √ λ12 dΩ(ˆ) q q |q| dΩ(ˆ) √ = . 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. p1 .

while the final state often arises from a series of 2-particle processes combined with the decay of an intermediate particle (resonance). A specific reaction channel starts contributing at the threshold value (Eq.17) (11. pz = mT sinh y.23) which are not independent as s + t + u = m2 + m2 + m2 + m2 . a c with q = λab /4s and q ′ = being 0 or 180 degrees.. (11. The initial state. however. Writing E = mT cosh y. there is usually a preferred direction.. the rapidity y is defined T T x y y= 1 ln 2 E + pz E − pz = ln E + pz mT = tanh−1 pz . Of course there are not only 2-particle channels. At high energies.18) (11. Instead of the scattering angle. dt = 2 qq ′ d cos θcm and obtain for the two-body phase space element √ q ′ dΩcm λcd dΩcm √ = (11. . Consider the process a + b → c + d. pT = (px . pd ) = 4π s 4π 8π s 4π dt dt √ √ . The variable s is always larger than a c b d the minimal value (ma + mb )2 . pc .21) (11. In inclusive measurements no particles are detected in the final state. The minimum and maximum values for t correspond to θcm CM = m2 + m2 − 2 Ea Ec ± 2 qq ′ a c = m2 a + m2 c (s + m2 − m2 )(s + m2 − m2 ) a c b d − ± 2s √ λab λcd . 2s (11.Scattering theory 104 Kinematics of 2 → 2 scattering processes π− + p The simplest scattering process is 2 particles in and 2 particles out.20) → .22) (11. py ).27) with m2 = m2 − p2 = m2 + p2 + p2 . in which one particle is detected.24) Using the relation between t and cos θcm it is straightforward to express dΩcm in dt. usually many particles are produced.25) dR(s. H1 + H2 → h + X. usually is a 2-particle state. Examples appear in → π− + p 0 (11. where the first is referred to as elastic channel and the set of all other channels as the inelastic channels.19) (11. → π +n → π+ + π− + n The various possibilities are referred to as different reaction channels. An often used set of invariants are the Mandelstam variables. = (11. 11.4). E .26) = 16π q s 8π λab Kinematics of inclusive hard scattering processes In high energy (hard) scattering processes. s = (pa + pb )2 = (pc + pd )2 t = (pa − pc ) = (pb − pd ) u = (pa − pd )2 = (pb − pc )2 2 2 (11. Consider the 1-particle inclusive case. for instance the momenta of incoming (colliding) hadrons and it is useful to use for the produced particle rapidity as a variable. 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 . tmax min λcd /4s. in exclusive measurements all particles are detected.

b Analiticity of the field theoretically calculated result implies Ma¯→¯ (st = t. The one-particle phase space in terms of these variables becomes dy d|pT |2 dφ dy d2 pT |p| dη d|pT |2 dφ d3 p = = = . t-channel and u-channel processes respectively. (11. ¯ b (11. uu = (pa − p¯)2 = s.31) .30) (11. u). tt = s. ut = u) = Mab→cd (s. u) with s = (p1 + p2 )2 = (p3 + p4 )2 . which means that rapidity distributions dN/dy maintain their shape. ¯ tt = (pa − p¯)2 = s. η = − ln(tan(θ/2)) = tanh−1 (cos θ). b ut = (pa − pd )2 = u. t = (p1 − p3 )2 = (p2 − p4 )2 .2 Crossing symmetry In the previous chapter. tu = s. These variables are precisely the Mandelstam variables for the s-channel process (ab → cd). t. t. For large energies and not too small scattering angles (θ ≫ 1/γ) y is approximately equal to the pseudo-rapidity η.Scattering theory 105 It is convenient because under a boost (along z) with velocity β it changes as y → y − tanh−1 β. (11.28) which (only involving angles) is easier to determine. ¯ while for the u-channel process (ad → c¯ one has b) su = (pa + pd )2 = u. Given a two-to-two scattering process ab → cd one can relate the processes b p p a 2 1 d p 4 _ c −p p 1 3 d p 4 _ d −p p a 1 4 _ b −p 2 p 3 c ab cd a _ ac −p 2 _ _ b bd _ ad p 3 _ c cb They are referred to as s-channel. t. c bd Mad→c¯(su = u. ¯ tu = (pa − pc )2 = t. uu = s) = Mab→cd (s. With the momenta defined as in the figures above one has for all these processes the same amplitude M (s.29) (2π)3 2E 2 (2π)3 16π 2 2π E 16π 2 2π with at high energies the factor |p|/E ≃ 1. For the t-channel process (a¯ → ¯ one has c bd) st = (pa + pc )2 = t. This is known as crossing symmetry. u). 11. u = (p1 − p4 )2 = (p2 − p3 )2 . 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.

while the flux of the beam particles a is trf flux(a) = ρa · va . shown below for the case of equal masses. . T where V and T indicate the volume and the time in which the experiment is performed. t and u. say b) the cross section σ(a + b → c + .32) σ= trf T · V ρa ρb va co . The proportionality factor has the dimension of area and is called the cross section. t. _ ac cos s= 0 _ bd cos θ t =1 θt =− 1 t u cos θ u = 1 sθ u=− s t=0 cos θ s = 1 u= 0 _ ad _ cb ab 1 cd 1 cos θs =− 11.Scattering theory 106 One can also phrase it in the following way: one has a single analytic function Mab→cd (s.3 Cross sections and lifetimes Scattering process For a scattering process a + b → c + . i.) is defined as the proportionality factor in Nc = σ(a + b → c + . . 1 N . . (consider for convenience the rest frame for the target. . which for a density ρb is given by Nb = ρb · V . This is a consequence of the constraint 2 2 2 2 s + t + u = Ma + Mb + Mc + Md . Nc /T indicates the number of particles c detected in the scattering process. To see this one can make a two-dimensional plot for the variables s. λs λs ab cd 1 2 2 2 2 t(s − u) + (Ma − Mc )(Mb − Md ) . . Nb indicates the number of (target) particles b. To find the physical regions one looks for the boundaries of cos θs = cos θt = cos θu = 1 2 2 2 2 s(t − u) + (Ma − Mb )(Mc − Md ) . t λt λac bd 1 2 2 2 2 u(t − s) + (Ma − Md )(Mc − Mb ) . u) that represents physical amplitudes in the physical regions for three scattering processes.) · Nb · flux(a).e. u λu λad cb This defines the boundaries of the physical regions. . (11.

pn ).37) 2 ma T · V Fermi’s Golden Rule In both the scattering cross section and the decay constant the quantity N/T V appears.Scattering theory 107 Although this at first sight does not look covariant. the amount of decaying particles is proportional to the number of particles with proportionality factor the em decay width Γ.e. From the solution N (t) = N (0) e−Γ t one knows that the decay time τ = 1/Γ. T · V ρa (11. .39) One of the δ functions can be rewritten as T · V (remember the normalization of plane waves). N and T · V are covariant. the number of scattered or decayed particles is given by 2 N = (2π)4 δ 4 (Pi − Pf ) iMf i dR(p1 . (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 . Microscopically one has Ndecay = Na · Γ T or Γ= N 1 .36) (11. it is. . . 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. 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 . dt (11. N 1 . . σ= √ 2 λab T · V (11.35) This quantity is not covariant. . (11.38) (in which we can calculate −iMf i using Feynman diagrams). For the (proper) decay width one thus has 1 N Γ0 = .T V. (2π)4 δ 4 (Pi − Pf ) 2 = (2π)4 δ 4 (Pi − Pf ) = (2π)4 δ 4 (Pi − Pf ) d4 x ei(Pi −Pf )·x V.T d4 x = V · T (2π)4 δ 4 (Pi − Pf ).33) Decay of particles For the decay of particle a one has macroscopically dN = −Γ N. as expected.34) i.

t) λab 4π dt. 16π 2 s (11. (11. pn ) |M |2 q 32π 2 m2 dΩ |M |2 . . dσ = √ 2 λab and for instance for two particles dσ = q ′ M (s. (11. p1 . pn ). ˆ ˆ Inserted in the unitarity condition for M . .44) This can be used to get the full expression for dσ/dt for eµ and e+ e− scattering. . for which one has dσ/dΩ = |f (E.47) M (s. see 11. . θ) = −8π s ℓ (in analogy with the expansion for f (E. . The amplitude −M /8π s is the one to be compared with the quantum mechanical scattering amplitude f (E. . The sign difference comes from the (conventional) sign in relation between S and quantummechanical and relativistic scattering amplitude. .43) dΩcm = π M (s. . . i.40) T ·V Combining this with the expressions for the width or the cross section one obtains for the decay width Γ = 1 2m = dR(m2 . 11. note the sign and cos θ = qi · qf ).25) −i Mf i − (M † )f i = − ∗ dΩ(ˆn )Mnf (q f . p1 . The result is N = |Mf i |2 dR(s. (11.Scattering theory 108 (Using normalized wave packets these somewhat ill-defined manipulations can be made more rigorous). (11.41) (11.42) 2−body decay The differential cross section (final state not integrated over) is given by 1 |Mf i |2 dR(s. θ)|2 . 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 . 8π s 2π 8π s .45) Since the amplitudes also depend on all momenta the full result for two-particle intermediate states is (in CM. (11. p1 .e.4 Unitarity condition (S † )f n Sni = δf i The unitarity of the S-matrix. . i M† M √ − √ 8π s 8π s = fi n dΩn ∗ Mnf qn Mni √ √ . θcm ) √ q 8π s 2 2 (11. q n ). √ (2ℓ + 1)Mℓ (s) Pℓ (cos θ). θ) in quantum mechanics.46) Partial wave expansion Often it is useful to make a partial wave expansion for the amplitude M (s. . θ). pn ). q f ). for which the √ amplitudes squared have been calculated in the previous chapter. θ) or M (q i . (M † )f n (2π)4 δ 4 (Pi − Pn ) Mni . q n ) q n qn √ Mni (q i . . respectively. implies for the scattering matrix M .

σinel = π q2 2 (2ℓ + 1) (1 − ηℓ ).48) If only one channel is present this simplifies to ∗ ∗ −i (Mℓ − Mℓ ) = 2 qMℓ Mℓ .51) and for the case that this is the only channel (purely elastic scattering.49) or Im Mℓ = q |Mℓ |2 . η = 1) the result σel = 4π q2 (2ℓ + 1) sin2 δℓ .54) ℓ . † −i (Mℓ )f i − (Mℓ )f i = 2 † (Mℓ )f n qn (Mℓ )ni .Scattering theory 109 we obtain LHS = −i while for the RHS use is made of ℓ † (2ℓ + 1) Pℓ (ˆi · qf ) (Mℓ )f i − (Mℓ )f i .53) The difference is the inelastic cross section.52) From the imaginary part of M (s. Using σ= dΩ q ′ M (s. n (11. σel = 4π ℓ (2ℓ + 1)|Mℓ (s)|2 (2ℓ + 1) ℓ = 4π q2 ηℓ e 2i δℓ − 1 2i 2 . which allows writing Mℓ (s) = e2i δℓ (s) − 1 Sℓ (s) − 1 = . θcm ) √ q 8π s 2 . the total cross section can be determined. Show that σT = = 4π q 2π q2 (2ℓ + 1) Im Mℓ (s) ℓ ℓ (2ℓ + 1) (1 − ηℓ cos 2δℓ ). (11. ℓ (11. (11. (11. 0).50) where Sℓ (s) satisfies |Sℓ (s)| = 1 and δℓ (s) is the phase shift. parametrized as Sℓ (s) = ηℓ (s) exp(2i δℓ (s)).e. in combination with the partial wave expansion for the amplitudes M and the orthogonality of the Legendre polynomials immediately gives for the elastic channel. q ˆ i. (11. 2i q 2i q (11. In general a given channel has |Sℓ (s)| ≤ 1. 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 quantity Γ is precisely the width for unstable particles.58) . it is straightforward to write down the partial wave amplitude. This is what we will do to discuss decay widths. 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. however. 11. (11. For instance the tree-level calculation of Feynman diagrams at order e2 was real. 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. For the amplitude in a scattering process going through a resonance. The product of amplitudes is of order e4 . . 0). 0) ∝ e−i(E−iΓ/2)t . . For an unstable particle one expects that U (t. which is given by a sum of the partial widths into the different channels. 0)|2 = e−Γt .57) This shows that Γ is the width of the resonance. This is achieved with a propagator i∆R (k) = i k 2 − M 2 + iM Γ (11. . pn )MRn (2π)4 δ 4 (k − Pn ) MnR = 2M n Γn = 2M Γ. such that |U (t. ∝ U (t. in which case.e. We note that unitarity is generally broken in a finite order calculation.e.56) (again disregarding spin). specifically one has for t > 0 that the Fourier transform is dk 0 e−ik t ∆(k) ∝ 0 dk 0 e−ik t (k 0 − Ek + iǫ)(k 0 + Ek − iǫ) 0 (t>0) ∝ e−iEk t .5 Unstable particles i − M 2 + iǫ For a stable particle the propagator is i∆(k) = k2 (11. To check unitarity we need the amplitude at order e4 . the time-evolution operator. 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 .55) (note that we have disregarded spin).Scattering theory 110 Note that the total cross section is maximal in the case of full absorption. σel = σinel . but still real. i. η = 0. The prescription for the pole structure. one has poles at k 0 = ±(Ek − iǫ) where Ek = + k2 + M 2 guarantees the correct behavior. The unitarity condition then states 2 Im ∆−1 (k) = R n † dR(p1 . . (q Mℓ )ij (s) = −M Γi Γj . s − M 2 + iM Γ (11. i.

26 keV. 1 cm . At resonance the cross section σT (π + p) is about 210 mb. u) = g(t. the full width is Γ = 88 keV. one can reconstruct the resonance shape for different numbers of neutrino species. cm |va − vb | in the center of mass system for two colliding particles a and b. Three famous resonances are: tan δℓ (s) = • The ∆-resonance seen in pion-nucleon scattering.49 GeV.1 Show that the cross section for electron-electron scattering (exercise 10. it is furthermore easy to prove that the cross section is given by σel = M 2 Γ2 4π (2ℓ + 1) . Its mass is M = 3096. Essentially this resonance can decay into quark-antiquark pairs or into pairs of charged leptons. where furthermore σinel = 0.e. Three neutrinos explain the resonance shape. Exercises Exercise 11. Note that because of the presence of a background the phase shift at resonance may actually be shifted. • The Z 0 resonance in e+ e− scattering with M = 91. This is a narrow resonance discovered in 1974. 2 q (s − M 2 )2 + M 2 Γ2 (11. The cross section at resonance is about 30 nb. All these decays can be seen and leave an ’invisible’ width of 498 MeV.2 GeV. a fast change in the phase shift for a narrow resonance. The cross section σT (π − p) also shows a resonance with the same width with a value of about 70 mb.60) M2 − s showing that the phase shift at resonance rises through δ = π/2 with a ’velocity’ ∂δ/∂s = 1/M Γ. the partial width into e+ e− is Γee = 5. Γ = 2. which is attributed to neutrinos. t)} . Its mass is M = 1232 MeV. This implies that the resonance has spin J = 3/2 (decaying in a P-wave (ℓ = 1) pion-nucleon state) and isospin I = 3/2 (the latter under the assumption that isospin is conserved for the strong interactions). its width Γ = 120 MeV. From this one sees that a resonance has the same shape in all channels but different strength.2 Calculate the flux factor. Limiting ourselves to a resonance in one channel.59) reaching the unitarity limit for s = M 2 . (11. u) + f (u. This characteristic shape of a resonance is called the Breit-Wigner shape.4) can be written as 4πα2 dσ = {f (t. u) + g(t. The phase shift in the resonating channel near the resonance is given by MΓ . Knowing that each neutrino contributes about 160 MeV (see next chapter). dt s(s − 4m2 ) f (t. The half-width of the resonance is M Γ. t) + g(u.88 MeV. u) = 1 1 2 (s + u2 ) + 4m2 (t − m2 ) t2 2 1 1 2 ( s − m2 )( s − 3m2 ) tu 2 2 with Exercise 11.Scattering theory 111 (Prove this using the unitarity condition for partial waves). ρcm ρcm a b . • The J/ψ resonance in e+ e− scattering. i.

999 877 Γ and Γ(π − → e− νe ) = 0.000 123 Γ. −M∆ Γ∆ /q . we have here already used unitarity to find a finite imaginary part. a quantity which we will ¯ ¯ encounter in the next chapter (section 12. As discussed for unstable particles.2). . 4.4 The ∆ resonance in N π scattering appears in one partial wave (a P-wave) as a pole at the position 2 s = M∆ − iM∆ Γ∆ . M1 (s) = 2 s − M∆ + iM∆Γ∆ when Γ∆ is approximately constant near this pole.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.Scattering theory 112 Exercise 11. Exercise 11.6 × 10−8 s and the relative decay widths Γ(π − → µ− νµ ) = 0. Show that unitarity can be used one step further to fix also the numerator of the amplitude near the pole. Calculate now fπ .8. Compare the results with the experimental π − lifetime τ = 2.

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. (12. one defines Dµ φ(x) ≡ ∂µ − ig W µ φ(x). ∂µ φ) the behavior of ∂µ φ under a local gauge transformation.1) with cabc known as the structure constants of the group. The extension to non-abelian gauge theories is straightforward. which satisfy commutation relations [Ta . (12. The photon field Aµ was introduced as to render the lagrangian invariant under local gauge transformations.2) where La is a representation matrix for the representation to which φ belongs. 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). In an abelian group the structure constants would be zero (for instance the trivial example of U (1)). 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). φ −→ U (θ) φ = ei θ·L φ = φ − θ × φ. 113 (12.5) Introducing as many gauge fields as there are generators in the group. φ(x) ∂µ φ(x) −→ −→ U (θ)φ(x). for a threecomponent field (written as φ) under an SO(3) or SU (2) symmetry transformation. U (θ) = ei θ (x)La . i.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.4) (12.3) The complication arises (as in the abelian case) when one considers for a lagrangian density a L (φ.6) .e. inf (12. which are conveniently combined a in the matrix valued field W µ = Wµ La . For a compact Lie-group they are antisymmetric in the three indices. U (θ)∂µ φ(x) + (∂µ U (θ)) φ(x). The symmetry group is a Lie-group G generated by generators Ta . Tb ] = i cabc Tc .

    ψb the wave function of the baryons (such as nucleons and deltas) form a singlet under SU (3)C . 2 . W ν ]. invariant under local SU (3)C transformations has proven to be a good candidate for the microscopic theory of the strong interactions. 2 ∗ λ − a for antiquarks. Dµ φ) − Ga Gµν a 2 4 µν (12.10) The gauge-invariant lagrangian density is now constructed as Gµν → U (θ) Gµν U −1 (θ).8) (12. Dµ φ) − Tr Gµν Gµν = L (φ. Generalizing the expression in terms of the covariant derivatives. i. g µ ν (12. g (12.The standard model 114 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). µν i [D . We have Gµν = with for the explicit fields transforming like a a b c Ga = ∂µ Wν − ∂ν Wµ + g cabc Wµ Wν . the necessity to have an antisymmetric wave function for the ∆++ particle consisting of three up quarks (each with charge +(2/3)e). the theory describing the interactions of the colored quarks. 1 1 L (φ. Note that the gauge fields must be massless. ∂µ φ) −→ L (φ.11) 1 with the standard normalization Tr(La Lb ) = 2 δab . The representation matrices for the quarks and antiquarks in the fundamental representation are given by Fa Fa = = λa for quarks. The existence of an extra degree of freedom for each species of quarks is evident for several reasons. For abelian fields Fµν = ∂µ Aν − ∂ν Aµ = (i/g)[Dµ . provides a gauge invariant definition for Gµν .9) (12. QCD.12) The nonabelian gauge theory that is obtained by making the ’free’ quark lagrangian. having three components in color space    ψ   r    ψ =  ψg  . Dν ] is gauge invariant. an example of a nonabelian gauge theory As an example of a nonabelian gauge theory consider quantum chromodynamics (QCD). 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. e. as a 2 a µa mass term ∝ MW Wµ W would break gauge invariance. With the quarks belonging to the fundamental (three-dimensional) representation of SU (3)C . D ] = ∂µ W ν − ∂ν W µ − ig [W µ . 1 |color = √ (|rgb − |grb + |gbr − |bgr + |brg − |rbg ) . for one specific species (flavor) of quarks just the Dirac lagrangian for an elementary fermion. ∂ L = i ψ/ ψ − m ψψ. In the nonabelian case ∂µ W a − ∂ν W a does not provide a gauge invariant candidate for Gµν = Ga La .g.7) 1 1 a ∂µ θa = Wµ + Dµ θa . 6 (12.e. µν ν µ as can be checked easily.

. 8 > > > λ5 = > > : i 9 −i > > > > > .1: The vectors belonging to internal space located at each point in (one-dimensional) space which satisfy commutation relations [Fa . Let fields Aa (x) determine the ’parallel displacements’ in the internal space. Ryder. forms a vector in the internal space (see fig. : : : 8 > > > λ4 = > > : 1 9 1 > > > > > . 8 > 1 > 1 > λ8 = √ > > 3: −2 . i. i. ν µ µ ν D ∂ A ∂ 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 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). chapter 7. For writing down the complete set of Feynman rules it is necessary to account for the gauge symmetry in the quantization procedure. connect µ 1 The Gell-Mann matrices are the eight traceless hermitean matrices generating SU (3) transformations. i. x ψ(x) denotes a field vector ψ(x) which belongs to a representation of G. The (locally) gauge invariant lagrangian density is 1 a L = − Fµν F µν a + i ψ/ ψ − m ψψ.g.1). 1 9 > > > 1 > > . D 4 with Dµ ψ = ∂µ ψ − ig Aa Fa ψ. µ (12. Consider the space x x G (called a fibre bundle).e. At each space-time point x there is considered to be a copy of an internal space G (say spin or isospin). 9 9 9 8 8 8 −i 1 > > > > 1 > > > > > > > > > > > > > i > 1 > > > −1 λ3 = > λ2 = > λ1 = > > > > > > > .) 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). 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 superscript x denotes that it is expressed with respect to the frame at point x. 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. 12. This will lead (depending on the choice of gauge conditions) to the presence of ghost fields. 1 8 > > > λ6 = > > : 9 > > > > > . .e. (For more details see e. In each of these spaces a reference frame is defined.13) a Fµν = ∂µ Aa − ∂ν Aa + g cabc Ab Ac . 8 > > > λ7 = > > : i 9 > > > −i > > . the basis of x G.e.The standard model xψ x+dx 115 (x) ψ(x) x+dx ψ(x+dx) x-space internal space Figure 12.

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 . x+dx ψ(x + dx) = 1 + dxµ Dµ x+dx ψ(x). The relation in Eq.14) (12. requiring that the ’connection’ Aµ (x) is such that Dµ → S(x)D µ S −1 (x). (12.17) The three equations given so far immediately give Dµ = ∂µ − ig Aµ (x). (12. What is this physical effect by which the ’connection’ Aµ can be observed? For this consider the phase around a closed loop.The standard model the basis for x G and x+dx 116 G. We note that local gauge invariance requires that we can modify all local systems with a (local) unitary transformation S(x). ds s dxµ ds′ P exp ig A (s′ ) ψ(0).18) or considering a path xµ (s) from a fixed origin (0) to point x. ds which is solved by dψ(s) ds = dxµ . which connects two identical vectors. A ’constant’ vector that only rotates because of the arbitrary definitions of local frames satisfies Dµ ψ(x) = 0.15) ψ(x). (12.e. x+dx ψ(x) = = 1 + ig dxµ Aa (x)Ta µ 1 + ig dx Aµ (x) µ x x ψ(x) (12. In principle such a phase in a given point is not observable. If there is a ’true’ difference in the vector ψ(x) and ψ(x + dx) it is denoted with the covariant derivative connecting the vectors expressed with respect to the same basis.19) This gives rise to a (path dependent) phase in each point. i. However.e. if two different paths to the same point give different phases the effects can be observed. (12. i. 12.16) which in the presence of the ’connection’ Aµ differs from the total change between x ψ(x) and x+dx ψ(x+ dx). Dν ] ψ(x) 1 − ig dσ µν Gµν ψ(x). = = . ds′ µ 0 ig Aµ (s) ψ(s) R dxµ Aµ (x) ψ(s) = which is the path-ordered integral denoted ψ(x) = P eig P ψ(0). -dx -dy µ µ dyµ µ dx For a vector. should be independent of such transformations. ∂µ − ig Aµ (x) ψ(x) = 0 dxµ ∂µ − ig Aµ (x(s)) ψ(x) = 0.16. x x+dx ψ(x + dx) = (1 + dxµ ∂µ ) ψ(x). but expresses them with respect to different bases.

    φ3 (12. the groundstate itself appears degenerate. 12. which means that by an appropriate transformation S(x) it can be gauged away (see the example of Aharonov-Bohm effect). 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). 12. Consider as an example a ferromagnet with an interaction hamiltonian of the form H=− Jij Si · Sj . As there can be one and only one groundstate. 2 2 4 −V (φ) The potential V (φ) is shown in fig. In this case there are many possible groundstates (determined by a fixed direction in space).g.2 Spontaneous symmetry breaking In this section we consider the situation that the groundstate of a physical system is degenerate. 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. e. i>j which is rotationally invariant.2. Only if this quantity is nonzero a physically observable effect of Aµ exists. Nature will choose one. Nevertheless. . If it is zero one has dxµ Aµ (x) = 0. 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 geometric language the effect on parallel transport of a vector depends on the ’curvature’ Gµν . in reality a not perfectly symmetric situation.e. Classically the (time-independent) ground state is found for V(φ ) F |φ| Figure 12.The standard model 117 where Gµν = (i/g)[Dµ .    φ1      φ =  φ2  . external magnetic fields.20) with a lagrangian density of the form L = 1 1 1 ∂µ φ · ∂ µ φ − m2 φ · φ − λ(φ · φ)2 . a theory for a scalar degree of freedom (a scalar field) having three (real) components. and equivalently Aµ can be considered as a pure gauge effect. This characterizes spontaneous symmetry breaking. Dν ]. 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. usually being (slightly) prejudiced by impurities. i.

In order to see the consequences for the particle spectrum of the theory we construct the lagrangian in terms of the fields ϕ1 . It is appropriate to redefine the field as    ϕ1     ϕ .e. has less symmetry. A quantum field theory has only one nondegenerate groundstate |0 . while the lagrangian including the nonzero groundstate expectation value chosen by nature. . The presence of a nonzero value for the classical groundstate value of the field will have an effect when the field is quantized.24) 1 2 2 4 1 1 1 = (∂µ ϕ1 )2 + (∂µ ϕ2 )2 + (∂µ η)2 + m2 η 2 + . therefore a choice must be made. In the case of degeneracy. (12. φ = ϕc + φquantum where for the (operator-valued) coefficients in the quantum field one wants 0|c† = c|0 = 0. while the quadratic terms must correspond with non-negative masses. It is sufficient to do this to second order in the fields as the higher (cubic.) terms constitute interaction terms. It is only invariant under rotations around the 3-axis. i. The field along the third axis plays a special role because of the choice of the vacuum expectation value. λ such that 0|ϕ1 |0 = 0|ϕ2 |0 = 0|η|0 = 0. In this case the spectrum is described . for the vacuum Qa |0 = 0. The situation now is the following. 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 .21) Stability of the action requires the classical groundstate ϕc to have a well-defined value (which can be nonzero). ϕ2 and η. (12. In this situation one speaks of spontaneous symmetry breaking. .  φ= (12. They are known as the Weyl mode or the Goldstone mode: Weyl mode. so one has 0|vecphiquantum |0 = 0 and hence 0|φ|0 = ϕc . The η massless particles are called Goldstone bosons.The standard model 118 a constant field (∇φ = 0) and the condition ∂V ∂φ ϕc =0 −→ ϕc · ϕc = 0 or ϕc · ϕc = − m2 ≡ F 2. .22)     F Realization of symmetries In this section we want to discuss a bit more formal the two possible ways that a symmetry can be implemented. say    0      0|φ|0 =  0  .23) 2     F +η the latter only forming a minimum for m2 < 0. The classical groundstate appears degenerate.25) 2 2 2 Therefore there are 2 massless scalar particles. Any constant field with ’length’ F is a possible groundstate. Writing the field φ as a sum of a classical and a quantum field. etc. In this mode the lagrangian and the vacuum are both invariant under a set of symmetry transformations generated by Qa . (12. 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 1 − m2 (F + η)2 − λ (ϕ2 + ϕ2 + F 2 + η 2 + 2 F η)2 (12. The original lagrangian contained an SO(3) invariance under (length conserving) rotations among the three fields.

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. and one must have mπa = 0. denoted |π a (k) . The conserved currents corresponding to this symmetry transformation are found directly using Noether’s theorem (see chapter 6).e. a ∂ µ Jµ = 0. (12.29) (Note that each of the entries in the vector for ψ is a 4-component Dirac spinor).      . (12.26) for all the states labeled by a corresponding to ’broken’ generators. Known examples are rotational symmetry and the fact that the spectrum shows multiplets labeled by angular momentum ℓ (with members labeled by m). . a massless Goldstone boson for each ’broken’ generator. there must be a nonzero expectation value 0|Jµ (x)|π (k) . down. π (12. This lagrangian density then is invariant under unitary (vector) transformations in the flavor space. called flavors. strange. a and a′ are degenerate states. [Qa . (12. The generators Qa (in that case the rotation operators Lz .  . This means that they are operators that create states from the vacuum. Goldstone mode. As the a generators for a symmetry are precisely the zero-components of a conserved current Jµ (x) integrated a a over space. H] = 0. etc. it is easy to see that one gets ∂µ V µ = i ψ [M. A bit more formal.e.   . V µ = ψγ µ T ψ. irrespective of the fact if they annihilate the vacuum. the lagrangian for the quarks consists of the free Dirac lagrangian for each of the types of quarks. T ] ψ. i. Taking the derivative.The standard model 119 by degenerate representations of the symmetry group.28) L = ψ(i/ − M )ψ. i. suggesting the stronger relation ∂ µ Jµ (x) = fπ m2 a π a (x). Neglecting at this point the local color symmetry.27) If the transformations in the lagrangian give rise to a symmetry the Noether currents are conserved.32) (12. Including a sum over the different flavors (up. In this mode the lagrangian is invariant but Qa |0 = 0 for a number of generators. a 0|∂ µ Jµ (x)|π b (k) = fπ k 2 ei k·x δab = fπ m2 a ei k·x δab . if the generators Qa generate a symmetry.     . π Chiral symmetry An example of spontaneous symmetry breaking is chiral symmetry breaking in QCD.. i. Using the Dirac equation. Note that for the fields π a (x) one would have a the relation 0|π a (x)|π a (k) = ei k·x .31) . T = τ /2. ψ −→ ei α·T ψ.30) which for instance including only two flavors form an SU (2)V symmetry (isospin symmetry) generated by the Pauli matrices. 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. where ψ is extended to a vector in flavor space and M is a diagonal matrix.      mu  ψu        ψ     md     M = ψ =  d .e.) one can write ∂ (12. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .

nature has choosen the path SU (2)L ⊗ SU (2)R =⇒ SU (2)V . with almost degenerate masses (Mp = 938. (12. in terms of which the Dirac lagrangian 2 density looks like ∂ ∂ (12.e. The Goldstone bosons of QCD are the pions for which fπ = 93 MeV and which have a mass of mπ ≈ 138 MeV/c2 . SU (2)V (isospin) symmetry is good if the up and down quark masses are identical. 12. i.e. but also a triplet (isospin 1) of pions. T } γ5 ψ. behaves as a pseudoscalar particle (spin zero. but the fact that the symmetry is not perfect and the right hand side of Eq. M = m · 1. parity would not play a role in the world. for which the field (according to the discussion above) has the same behavior under parity. I. Note that these transformations also work on the spinor indices. (leaving out the flavor structure) the same as ψγ5 ψ. much smaller than any of the other mesons or baryons. where the quark masses are not completely zero. a doublet (isospin 1/2) of nucleons. which is equivalent 2 to the V-A symmetry.e.34) In this case ∂µ Aµ = 0 will be true if the quarks have zero mass. it is easy to see that one gets ∂µ Aµ = i ψ {M.The standard model 120 Furthermore ∂µ V µ = 0 ⇐⇒ [M. nucleons and atomic nuclei has not only an isospin or vector symmetry SU (2)V but also an axial vector symmetry SU (2)A . T γ5 = τ γ5 /2. which is approximately true for the up and down quarks. is by nature not realized in the Weyl mode. i. as the quantity ∂µ Aµ (x). gives also rise to a nonzero mass for the Goldstone bosons according to Eq. chiral symmetry is not perfect. In the real world. socalled axial vector transformations.27. This symmetry is realized in the Weyl mode with the spectrum of QCD showing an almost perfect isospin symmetry. parity minus). Therefore realization in the Weyl mode would require that all particles come double with positive and negative parity. ψ −→ ei α·T γ5 ψ. if in flavor space M is proportional to the unit matrix. the lagrangian density is invariant under left (L) and right (R) rotations independently. both are very small. T ] = 0. The currents corresponding to this symmetry transformation are again found using Noether’s theorem. Still the basic fact that the generators acting on the vacuum give a nonzero result (i. etc. 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.35 is nonzero. Aµ = ψγ µ T γ5 ψ. The chiral fields ψR and ψL are transformed into each other under parity. etc. Therefore the world of up and down quarks describing pions. while the groundstate is only invariant under isospin rotations (R = L).e. 12. Using the Dirac equation. From the number of broken generators it is clear that one expects three massless Goldstone bosons. . is what happens in the real world. There exists another set of symmetry transformations. (12. or. We know that mesons and baryons (such as the nucleons) have a well-defined parity that is conserved. generated by TR/L = 1 (1 ± γ5 )T . ψR/L = 1 (1 ± γ5 )ψ.g.33) which for instance including only two flavors form SU (2)A transformations generated by the Pauli matrices. That the massless theory has this symmetry can also be seen by writing it down for the socalled lefthanded and righthanded fermions. 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. This combined symmetry is what one calls chiral symmetry.e. proton and neutron. This situation.35) (12. This symmetry. e.36) L = i ψL / ψL + i ψR / ψR − ψR M ψL − ψL M ψR .6 MeV/c2 ). How can we see this. stated equivalently. fπ = 0 remains. i.3 MeV/c2 and Mn = 939. From group theory (Schur’s theorem) one knows that the latter can only be true. however.

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.e. thus 9 independent degrees of freedom. We have the possibility to perform local gauge transformations. 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). (12.    The symmetry is broken in the same way as before and the same choice for the vacuum.     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 µ ν ν µ 1 2 = − (∂µ Wν − ∂ν Wµ ) · (∂ W − ∂ W ) − g F Wµ W µ 1 + Wµ W µ 2 4 2 1 + (∂µ η)2 + m2 η 2 + . Dµ φ = Gµν = ∂µ φ + g Wµ × φ. (12. 2 massive vector fields or spin 1 bosons (2 × 3) and one scalar field (1). 1 1 L = − Gµν · Gµν + Dµ φ · Dµ φ − V (φ).40) a Dµ φ = ∂µ φ − ig Wµ La φ. Initially there are 3 massless gauge fields (each. The difference comes when we reparametrize the field φ. made into a gauge theory. One may wonder about the degrees of freedom.The standard model 121 12. in this case three-dimensional) representation reads (La )ij = −i ǫaij . having two independent spin components) and three scalar fields (one degree of freedom each). .42) 2 from which one reads off that the particle content of the theory consists of one massless gauge boson 3 1 2 (Wµ ).     0 φ= 0 = (12. The latter is a spin 0 particle (real scalar field) called a Higgs particle. . like a photon. two massive bosons (Wµ and Wµ with MW = gF ) and a massive scalar particle (η with m2 = η 2 −2 m .39) (12. while the number of massive gauge bosons coincides with the number of ’broken’ generators. one sees that the fields Wµ and Gµν also can be represented as three-component fields. is made. (12. 4 2 where Since the explicit (adjoint.    0      ϕc = 0|φ|0 =  0  . Considering the same example with rotational symmetry (SO(3)) as for spontaneous symmetry breaking of a scalar field (Higgs field) with three components.38) ∂µ Wν − ∂ν Wµ + g Wµ × Wν . but one has 1 massless gauge field (2). Therefore we can always rotate the field φ into the 3-direction in order to simplify the calculation. . . again 9 degrees of freedom.     0  0        .41)         φ3 F +η Explicitly one then has  2 2  gF Wµ + g Wµ η   −gF W 1 − g W 1 η Dµ φ = ∂µ φ + g Wµ × φ =   µ µ  ∂µ η    . After symmetry breaking the same number (as expected) comes out.37) (12. i. as in this case there are no massless Goldstone bosons.

νµ and ντ ) and their antiparticles. (12.The standard model 122 12. neutrinos. Thus L (f ) = i eR / eR + i eL / eL + i νeL / νeL + (µ. = ∂µ L − (12.48) (12. ∂ ∂ ∂ (12. U(1)Y U(1)Y (12. 1 As they are massless. however.50) (12. τ ). under which the lefthanded and righthanded particles with YW (L) = −1 and YW (R) = −2 transform with eiβYW /2 .44) which has an SU (2)W symmetry under transformations eiα·TW . that gives a good description of the physical world. R −→ e−i β R. left. electromagnetic and weak forces. etc. (12. implies certain nontrivial relations between masses and mixing angles that can be tested experimentally and sofar are in excellent agreement with experiment. 2 (12. 2 2 = ∂µ R + i g ′ Bµ R. ∂ ∂ = i L/ L + i R/ R. explicitly L −→ e−i β/2 L.) and the gauge bosons responsible for the strong.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. In the standard model one starts with a very simple basic lagrangian for (massless) fermions which exhibits more symmetry than observed in nature.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     1 3 and TW = 2 +1/2 −1/2 with TW = R = eR L (f ) 3 with TW = 0 and TW = 0. muons. i.). ψR/L = 2 (1 ± γ5 )ψ decouple. explicitly L R SU(2)W SU(2)W −→ ei α·τ /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. By introducing gauge fields and breaking the symmetry a more complex lagrangian is obtained. the leptons (elektrons. explicitly Dµ L Dµ R i i g Wµ · τ L + g ′ Bµ L.51) .   ν L= e  eL and Thus the lagrangian density is symmetry under which eR forms a singlet. its corresponding neutrino (νe . muon µ− or tau τ − ). The lagrangian for the leptons consists of three families each containing an elementary fermion (electron e− . the quarks (up. For the neutrino only a lefthanded particle (and righthanded antiparticle) exist. R.47) and YW is considered as an operator that generates a U (1)Y symmetry. etc.45) (12.e. The procedure.43) One introduces a (weak) SU (2)W particles form a doublet.and righthanded particles. down. This is written as 3 Q = TW + YW .

2 and 3 may be eliminated (the necessary SU (2)W rotation is precisely e−iθ(x)·τ ). leading to the parametrization   1  0    (12. (12. −V (φ) L and (h2) = −Ge (LφR + Rφ L). the U (1)Q symmetry (generated by the charge operator). . this leads to the lagrangian L (h1)     .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 2 − (∂µ Wν − ∂ν Wµ + g Wµ × Wν ) − (∂µ Bν − ∂ν Bµ )2 . . TW = ±1 or 0 and YW = 0 (thus Q = TW ) and 3 Bµ is a singlet under SU (2)W (TW = TW = 0) and also has YW = 0.58) (12.53) φ= 0 = √   1  φ (θ4 − iθ3 )  2 with TW = 1/2 and YW = 1 is introduced. † i i g Wµ · τ − g ′ Bµ )φ. which assures with the choice of YW of the Higgs field assures that Q generates the remaining U (1) symmetry. Using local gauge invariance θi for i = 1. .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. + 8 (12.57) = 1 g2 v2 1 2 (Wµ )2 + (Wµ )2 (∂µ h)2 + m2 h2 + 2 8 v2 2 3 + gWµ − g ′ Bµ + . a complex Higgs field   +   1  φ   √2 (θ2 + iθ1 )     (12.54) = (Dµ φ)† (Dµ φ) −m2 φ† φ − λ (φ† φ)2 . Putting this in leads to L (f ) =L (f 1) +L (f 2) . 4 4 In order to break the symmetry to the symmetry of the physical world.56) φ(x) = √   v + h(x) 2 Dµ φ = (∂µ − and     Dµ φ =     = ig 2 1 √ 2 ∂µ h 1 2 Wµ −iWµ √ (v + 2 3 gWµ −g′ Bµ i √ −2 2 h) (v + h) Up to cubic terms. (12. 8 g2 v2 1 + − (∂µ h)2 + m2 h2 + (Wµ )2 + (Wµ )2 2 8 (g 2 + g ′2 ) v 2 (Zµ )2 + .The standard model 123 3 3 where Wµ is a triplet of gauge bosons with TW = 1. . (12.59) . with the following lagrangian density consisting of a symmetry breaking piece and a coupling to the fermions. For the classical field the choice θ4 = v is made.    (12. . L where L (h1) (h) =L (h1) +L (h2) .

2 + g ′2 g (12. 2 Zµ (12. .60) (12.66) From the coupling to the photon.63) (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 1 1 g sin2 θW eR γ µ eR − cos 2θW eL γ µ eL + νe γ µ νe + cos θW 2 2 g µ − µ + + √ νe γ eL Wµ + eL γ νe Wµ .68) g2 ρ 5 2 (νµ γρ (1 − γ5 )µ) (eγ (1 − γ )νe ) 8 MW (jL (µ)† )ρ (jL )ρ (e) GF (µ)† (e) ≡ i √ (jL )ρ (jL )ρ .67) The coupling of electrons or muons to their respective neutrinos. σ (νµ γ ρ µL ) 2 2 (eL γ νe ) 2 k + MW (12.62) 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 . for instance in the amplitude for the decay of the muon νµ µ− W − e− νµ = µ− e− νe νe is given by −i M = − ≈ i g2 −i gρσ + .64) (12. 2 3 g Wµ − g ′ Bµ 3 ≡ cos θW Wµ − sin θW Bµ .69) . 4 cos2 θW cos2 θW 0. . (12.The standard model 124 where the quadratically appearing gauge fields that are furthermore eigenstates of the charge operator are ± Wµ = = = Zµ Aµ 1 1 2 √ Wµ ± i Wµ . 4 2 2 2 MW g v = . 2 (12.65) The weak mixing angle is related to the ratio of coupling constants. (12. g ′ /g = tan θW . g 2 + g ′2 3 g ′ Wµ + g Bµ 3 ≡ sin θW Wµ + cos θW Bµ . we can read off e = g sin θW = g ′ cos θW .61) (12.

g2 MZ 2 2 (CV + CA ). Γinvisible = 480 MeV one sees that three families of (light) neutrinos are allowed. = 0.The standard model 125 the good old four-point interaction introduced by Fermi to explain the weak interactions. From this coupling it is straightforward to calculate the partial width for Z 0 into a fermion-antifermion pair. 2 2 2 CV CA 3 = TW − 2 sin2 θW Q. = 80. 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 . from the MW mass is about 250 GeV.79) Γ(Z 0 → f f ) = 48π cos2 θW For the electron. For each neutrino species (with CV = 1/2 and CA = 1/2 one expects Γ(νν) ≈ 155 MeV.e. leptons with CV = −1/2 + 2 sin2 θW ≈ −0.166 4 × 10−5 GeV−2 . Finally we want to say something about the weak properties of the quarks.72) (12. With the choosen vacuum expectation value for the Higgs field. The coupling to the Higgs particle is weak as the value for v calculated e. (12. (12. i.77) (12.73) (12. such as e2 /4π GF 2 sin θW MW MZ ≈ 1/137.5 MeV (exp. g ′ and v determine a number of experimentally measurable quantities. (12.76) with (12. Comparing this with the total width into (invisible!) channels. i.19 GeV. muon or tau. (12. L (h2) u u d W eνe n −→ pe− νe ⇐⇒ d −→ ue− νe . The masses of the fermions and the coupling to the Higgs particle are contained in L (h2) . = 91. one obtains Ge Ge v = − √ (eL eR + eR eL ) − √ (eL eR + eR eL ) h 2 2 me eeh.80) = −me ee − v First.231 2.e.78) = 3 TW . the mass of the electron comes from the spontaneous symmetry breaking but is not predicted (it is in the coupling Ge ).75) = 1. me /v is extremely small. s W eνe Λ −→ pe− νe ¯ ⇐⇒ s −→ ue− νe .71) (12. Γe ≈ Γµ ≈ Γτ ≈ 83 MeV). (12.40 GeV. as appear for instance in the decay of the neutron or the decay of the Λ (quark content uds). one has the relation g2 e2 1 GF √ = .70) = = 2 2 sin2 θ 8 MW 2 v2 8 MW 2 W In this way the parameters g.74) (12. Actually including corrections corresponding to higher order diagrams the agreement for the decay width into electrons can be calculated much more accurately and the number of allowed (light) neutrinos turns to be even closer to three.g. ¯ .05 and CA = −1/2 we calculate Γ(e+ e− ) ≈ 78.

allowing us to write † Λu = Vu Gu Wu (12.82) space.q) = −QL φΛd DR − DR Λ† φ† QL − QL φc Λu UR − UR Λ† φc† QL u d (12. Decay of B-mesons containing b-quarks allow estimate of Vub .  ′   ′        b s d L L L   ′   Vud  d    ′     s  =  Vcd       ′  b Vtd L Vus Vcs Vts Vub Vcb Vtb           d      s       b L (12.  φ =    −φ− 0 2 For the (squared) complex matrices we can find positive eigenvalues. u u where we include now the three lefthanded quark doublets in QL . Comparing neutron decay and Λ decay one can get an estimate of the mixing parameter Vus in the socalled Cabibbo-Kobayashi-Maskawa mixing matrix. for which we need the appropriate weak isospin doublet  0∗    1   φ  c   = √  v +h .g. The imaginary part i η gives rise to CP violation in decays of K and B-mesons (containing s and b quarks. 12. The Λu and Λd are complex matrices in the 3×3 family and Λd Λ† = Vd G2 Vd† . A complication arises as it are not the ’mass’ eigenstates that appear in the weak isospin doublets but linear combinations of them. UR = uR cR tR . Thus one has L (h2.85) .83) where Vu and Vd are unitary matrices. In principle one complex phase is allowed in the most general form of the CKM matrix. respectively). A ≈ 0. each of these containing   the three families. CP violation requires three generations. Note that we need the conjugate Higgs field to get a U (1)Y singlet in the case of the charge +2/3 quarks.q) † † † =⇒ −DL Vd Md Wd DR − DR Wd Md Vd† DL − UL Vu Mu Wu UR − UR Wu Mu Vu UL (12. i.e. d d † and Λd = Vd Gd Wd . This is only true if the mixing matrix is at least three-dimensional.81) 3 This mixing allows all quarks with TW = −1/2 to decay into an up quark. † Λu Λ† = Vu G2 Vu . which can account for the (observed) CP violation of the weak interactions. e.5 Family mixing in the Higgs sector and neutrino masses The quark sector Allowing for the most general (Dirac) mass generating term in the lagrangian one starts with L (h2. (12. The magnitudes of the entries in the CKM matrix are nicely represented using the socalled Wolfenstein parametrization   λ λ3 A(ρ − i η)  1 − 1 λ2  2      + O(λ4 ) −λ 1 − 1 λ2 λ2 A V =   2   3 2 λ A(1 − ρ − i η) −λ A 1 with λ ≈ 0.22 and η ≈ 0.84) with Gu and Gd being real and positive and Wu and Wd being different unitary matrices.227.The standard model 126 where The quarks also turn out to fit into doublets of SU (2)W for the lefthanded species and into singlets for the righthanded quarks.34. the three righthanded quarks with charge +2/3 in UR and the three righthanded quarks with charges −1/3 in DR . etc. The Higgs field is still limited to one complex doublet.        t   c   u   ′  . but with different strength.82 and ρ ≈ 0.

ℓ) = −LφΛe ER − ER Λ† φ† L − Lφc Λn NR − NR Λ† φc† L e n (12.     0 0  c13 0 s13 eiδ   c12 s12 0   1        −s    0 1 0 c12 0  . 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. i. NL = Vn NL and NR = Wn NR µ mass γ µ V † V N mass . One then reads off that starting with the family basis as defined via the left doublets that the mass eigenstates (and states coupling to the Higgs field) involve the righthanded † mass † mass mass † mass states UR = Wu UR and DR = Wd DR and the lefthanded states UL = Vu UL and DL = † Vd DL . The mass fields ER † mass † mass = We ER . (12. The lepton sector (massive Dirac neutrinos) In principle a massive Dirac neutrino could be accounted for by a lagrangian of the type L (h2. since the weak current is the only place where they show up. and obtain L (h2.91) † with UPMNS = Vn Ve known as the Pontecorvo-Maki-Nakagawa-Sakata mixing matrix.92) 12         iδ 0 −s23 c23 0 0 1 −s13 e 0 c13 .The standard model 127 √ √ with Mu = Gu v/ 2 (diagonal matrix containing mu .87) is a weak doublet containing the three families of neutrinos (NL ) and charged leptons (EL ) and ER is † a three-family weak singlet. there is no family mixing for massless neutrinos. mc and mt ) and Md = Gd v/ 2 (diagonal matrix containing md . (12. The W -current then becomes mass mass E L γ µ NL = E L γ µ NL .89) with three righthanded neutrinos added to the previous case. the weak mass eigenstates are ′ weak † mass mass DL = DL = Vu Vd DL = VCKM DL . ms and mb ). i.86) the unitary CKM-matrix introduced above in an ad hoc way. mµ and mτ . we find massless neutrinos. decoupling from all known interactions. For the (massless) neutrino fields we just can redefine fields into NL = Ve† NL . As before. (12.ℓ) =⇒ −Me EL Ve We ER − ER We Ve† EL . one can write Λe = Ve Ge We and we find † L (h2. The lepton sector (massless neutrinos) For a lepton sector with a lagrangian density of the form L in which (h2. EL = Ve EL . e    N  L= L    EL (12.e.ℓ) = −LφΛe ER − ER Λ† φ† L. Working with the mass eigenstates one simply sees that the weak current coupling to the W ± mass † mass becomes U L γ µ DL = U L γ µ Vu Vd DL .e. 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 .ℓ) † † † =⇒ −EL Ve Me We ER − ER We Me Ve† EL − NL Vn Mn Wn NR − NR Wn Mn Vn NL . EL = Ve† EL . † † Again we continue as before now with matrices Λe = Ve Ge We and Λn = Vn Gn Wn . UPMNS =  0 c23 s23   (12.88) √ mass with Me = Ge v/ 2 the diagonal mass matrix with masses me .90) mass † mass mass † mass † We note that there are mass fields ER = We ER . (12.

Thus (disregarding family structure) one has two Majorana neutrinos.ν = − MR NR NR + MR NR NR . Actually. it is particularly useful because θ12 is essentially determined by solar neutrino oscillations requiring ∆m2 ≈ 8 × 10−5 12 eV2 (convention m2 > m1 ). coupled by a Dirac mass term. For the leptons. For the charged leptons there cannot exist a Majorana mass term as this would break the U(1) electromagnetic symmetry. Absorption of phases in the states is not possible for Majorana neutrinos. with for the righthanded sector a mass term MR . The way to circumvent this is to introduce as in the previous section righthanded neutrinos.and righthanded species then just form a Dirac fermion. the left. L mass. Thus 2 Υ ′ = nc + nL . couple the right.and lefthanded species through Dirac mass terms coming from the coupling to the Higgs sector as in the previous section. 1 c ∗ c L mass. This is called the see-saw mechanism (outlined below). Υ′ and Υ′ (satisfying Υc = Υ and as discussed in chapter 6.88 a Majorana mass term for the (lefthanded) neutrino mass eigenstates. a unitary matrix relating them to the weak eigenstates. 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 see-saw mechanism Consider (for one family N = n) the most general Lorentz invariant mass term for two independent Majorana spinors. (12. If the Majorana mass MR ≫ MD one actually obtains in a natural way one Majorana neutrino with a very small mass.5×10−3 eV2 . is to add in Eq.       0 0 1 containing three (CP-violating) phases (α1 . are equivalent to two decoupled Majorana neutrinos (see below). 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  . one must for the neutrinos as well as charged leptons.      2/3 0  =  − 1/6           1/2 1/2 0 0 1 0 1/3 1/2 − 1/6 (12. 12.96) VPMNS = UPMNS K with K =  0 eiα2 /2 0  . one being massive. hence the mixing matrix becomes   iα /2 0 0   e 1     (12. as above. while θ23 then is determined by atmospheric neutrino oscillations requiring |∆m2 | ≈ 2. however.94) although this option is not attractive as it violates the electroweak symmetry. For these light Majorana neutrinos one has.97) . 1 L Υ ′ = nR + nc .93) The lepton sector (massive Majorana fields) An even simpler option than sterile righthanded Dirac neutrinos.95) 2 In order to have more than a completely decoupled sector. Υc ≡ (ΥL )c = ΥR 2 1 L and Υc = ΥL ).ν =− 1 ∗ c c M L NL NL + M L NL NL .The standard model 128 a parametrization that in principle also could have been used for quarks. For the neutrino sector. (12. the massless and massive Majorana neutrinos. α2 and δ). We use here the primes starting with the weak eigenstates. 2 R ψ = nR + nL . In this case. 2 (12.

2 with real masses Mi . i. (12. This is a mixing problem with a symmetric (complex) mass matrix leading to two (real) mass eigenstates. and |a and |a′ belong to the same multiplet (there is a Qa such that |a′ = Qa |a ) then H|a = Ea |a implies that H|a′ = Ea |a′ .105) for i = 1.101)   . a and a′ are degenerate states.98) (12.e.1 Consider the case of the Weyl mode for symmetries. Exercises Exercise 12. For the situation ML = 0 and MR ≫ MD (taking MD real) one 2 finds M1 ≈ MD /MR and M2 ≈ MR . The mass matrix can be written as   ML |MD | eiφ     M = (12. [Qa .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)) . Prove that if the generators Qa generate a symmetry. H] = 0.100)  |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.103) for each of which one finds the lagrangians L = and we are left with two decoupled Majorana fields      Υ1L    = U ∗  nc  . i.104) ↔ 1 1 Υi i / Υi − Mi Υi Υi ∂ 4 2 (12. which implies U ∗ M † U † = U ∗ M ∗ U † = M0 and a ’normal’ diagonalization of the (hermitean) matrix M M † .The standard model 129 As the most general mass term in the lagrangian density we have L mass = − = taking ML and MR real and non-negative.e. This choice is possible without loss of generality because the phases can be absorbed into Υ′ and Υ′ (real must be replaced by hermitean if one includes 1 2 families). The diagonalization is analogous to what was done for the Λ-matrices and one finds U M U T = M0 with a (unitary) matrix U .c. which for MD = 0 is a pure Majorana lagrangian and for ML = MR = 0 and real MD represents the Dirac case. . related via    c   Υ1R    = U  nL  .  −    nR MD MR 2 (12.       Υ2R nR (12.99) (12.  L      Υ2L nR Υ1 and Υ2 . 2 U (M M † ) U † = M0 .  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.

4 Calculate the lifetime τ = 1/Γ for the top quark (t) assuming that the dominant decay mode is t → b + W +. mb ≈ 5 GeV and MW ≈ 80 GeV. 3 TW . Neglect the masses of fermions and use the fact that the sum over polarizations is 3 pµ pν ǫ(λ) (p)ǫ(λ)∗ (p) = −gµν + . The mass of the Z 0 is MZ = 91 GeV. (b) Calculate the width to electron-positron pair. i ψ {M. V µ = ψγ µ T ψ and Aµ = ψγ µ γ5 T ψ ∂µ V µ ∂µ A µ = = i ψ [M. at least up to a high (which?) order in λ. µ ν M2 λ=1 ¯ to calculate the width Γ(Z → f f ). T ] ψ. Exercise 12.2 Derive for the vector and axial vector currents.6 Check that the use of the Wolfenstein parametrization in the CKM matrix indeed gives a unitary matrix. Exercise 12.The standard model 130 Exercise 12. T } γ5 ψ.3 (a) The coupling of the Z 0 particle to fermions is described by the vertex −i with CV CA = = 3 TW − 2Q sin2 θW . g 2 cos θW f f CV γ µ − CA γ µ γ5 . ZZh. 0 MZ g2 ¯ Γ(Z 0 → f f ) = 48π cos2 θW f f CV 2 + CA 2 . In the standard model this coupling is described by the vertex −i g µ √ (γ − γ µ γ5 ) . Exercise 12.5 Show that the coupling to the Higgs (W + W − h. Write down the matrix element squared (averaged over initial spins and summed over final spins) for the decay of the Z 0 . Exercise 12. . the weak mixing angle is given by ¯ sin2 θW = 0. Γ(Z 0 → νe νe ). hhh and e+ e− h) are proportional to the mass squared (bosons) or mass (fermions) of the particles.231. 2 2 The masses are mt ≈ 175 GeV. and the width to a pair of neutrino’s. Note that you can find the answer without explicit construction of the interaction terms in the lagrangian. Γ(Z 0 → e+ e− ).

05 eV. Calculate the eigenvalues ML = 0 and MR = MX . Given that neutrino masses are of the order of 0. .The standard model 131 Exercise 12. Give the explicit expressions that relate ψ and ψ c with Υ1 and Υ2 .7 In this exercise two limits are investigated for the two-Majorana case. 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. what is the mass MX if we take for MD the electroweak symmetry breaking scale v (about 250 GeV). (a) Calculate for the special choice ML = MR = 0 and MD real. which leads to the socalled see-saw mechanism. (b) A more interesting situation is 0 = ML < |MD | ≪ MR .

Sign up to vote on this title
UsefulNot useful