You are on page 1of 220

arXiv:0906.1271v2 [physics.

ed-ph] 18 Dec 2009

Lessons in Particle Physics
Luis Anchordoqui and Francis Halzen
University of Wisconsin

©

2009

2

Contents
1 General Principles
1.1 Particle Zoo . . . . . . . . . . . . .
1.2 Canonical Quantization . . . . . . .
1.3 Lorentz Group . . . . . . . . . . . .
1.4 Klein-Gordon Equation . . . . . . .
1.5 Dirac Equation . . . . . . . . . . .
1.6 Nonrelativistic Perturbation Theory
2 Symmetries and Invariants
2.1 Noether Theorem . . . . . . . . . .
2.2 Gauge Invariance . . . . . . . . . .
2.2.1 Maxwell-Dirac Lagrangian .
2.2.2 Yang-Mills Lagrangian . . .
2.2.3 Isospin . . . . . . . . . . . .
2.3 Higgs Mechanism . . . . . . . . . .
2.4 Standard Model of Particle Physics
3 QED
3.1 Invariant Amplitude . . .
3.2 Unpolarized Cross Section
3.3 Mandelstam Variables . .
3.4 Feynman Rules . . . . . .
3.5 Beyond the Trees . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

5
5
9
11
15
18
34

.
.
.
.
.
.
.

39
39
42
42
45
49
52
59

.
.
.
.
.

67
67
76
79
86
92

4 Hard Scattering Processes
101
4.1 Deep Inelastic Scattering . . . . . . . . . . . . . . . . . . . . . 101
3

4.2 Parton Model . . . . . . . . . . . . . . . .
4.3 QCD Improved Parton Model . . . . . . .
4.4 Physics of Hadronic Jets . . . . . . . . . .
4.4.1 Hadroproduction of Direct Photons
4.4.2 Two-Jet Final States . . . . . . . .
5 Precision Electroweak Physics
5.1 Charged and Neutral Currents . . . . . .
5.2 Quark Flavor Mixing . . . . . . . . . . .
5.3 Scalars were already part of the Theory!
5.4 Electroweak Model @ Born Level . . . .
5.4.1 Interference in e+ e− annihilation
5.4.2 The NuTeV anomaly . . . . . . .
5.5 Radiative Corrections . . . . . . . . . . .
5.6 Lepton Flavor Mixing . . . . . . . . . . .
5.6.1 Neutrino Oscillations . . . . . . .
5.6.2 How to kill a vampire . . . . . . .
5.7 Closure . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

112
117
125
126
131

.
.
.
.
.
.
.
.
.
.
.

139
. 139
. 148
. 154
. 156
. 158
. 162
. 164
. 175
. 175
. 181
. 186

6 Dark Side of the Universe

189

A Decay Rate in Terms of M

193

B Trace Theorems and Properties of γ-Matrices

195

C Dimensional Regularization

199

D Mott Scattering

203

E Laboratory Kinematics

207

F

211

Spin- and Color-Averaged Cross Sections

G Monojets

213

H Muon Decay

215

I

219

Asymmetries at the Z-pole
4

it seems opportune to review our present understanding of particle interactions. the structure of matter and the forces by which it interacts. electromagnetic. Counter-circulating proton √ beams into head-on collisions at a center-of-mass energy s = 14 TeV. strong. These particles interact via four known basic forces – gravitational. Though gravity is the most obvious force in daily life. the strong force is mediated by gluons. the LHC will probe deeply into the sub-fermi distances. The Large Hadron Collider (LHC) promises to take the next leap in this direction.Chapter 1 General Principles 1. and weak – that can be characterized on the basis of the following four criteria: the types of particles that experience the force. opening a new territory where groundbreaking discoveries are expected. In the past half-century colossal strides were made in bringing Quantum Field Theory (QFT) to bear upon a wide variety of phenomena. the accepted model for elementary particle physics views quarks and leptons as the basic constituents of ordinary matter. at the deepest level. the W ± and Z 0 bosons transmit the weak force. the range over which the force is effective. Today. on 5 . In this spirit then. the relative strength of the force.1 Particle Zoo High energy physics seeks to understand. A comparison of the (approximate) relative force strengths is given in Table 1. The electromagnetic force is carried by the photon. and the nature of the particles that mediate the force. and the quantum of the gravitational force is called the graviton.1.

known as flavors: up (symbol: u).Table 1.2. strangeness.) Quarks are fermions with spin. bottom (b). charm c. bottomness. Type Relative Strength Field Particle Strong 1 gluons −2 Electromagnetic 10 photon −6 Weak 10 W ± Z0 Gravitational 10−38 graviton Table 1. and top (t). (Antiquarks have opposite signs of electric charge. strange (s). charm. “beauty” or bottomness b.21 and therefore should obey the exclusion principle. baryon number. down (d). ∆− = ddd. and Ω− = sss). Yet for three particular baryons (∆++ = uuu. and at least two quarks have their spin in the same direction because there are only 6 .12 strongly interacting objects which are known to form the composites collectively called hadrons: ( q q¯ (quark + antiquark) mesons integral spin → Bose statistics qqq (three quarks) baryons half-integral spin → Fermi statistics . baryon number B. The quarks are fractionally charged spin. their properties are given in Table 1. all three quarks would have the same quantum numbers.1: Relative strength of the four forces for two protons inside a nucleus. strangeness S. name symbol Q B S c b t 1 2 0 0 0 0 up u 3 3 1 1 0 0 0 0 down d −3 3 1 1 −1 0 0 0 strange s −3 3 1 2 0 1 0 0 charm c 3 3 1 1 bottom b −3 0 0 −1 0 3 1 1 0 0 0 1 top t −3 3 a nuclear scale it is the weakest of the four forces and its effect at the particle level can nearly always be ignored.2: Quark quantum numbers: charge Q. and topness. charm (c). There are six different types of quarks. and “truth” or topness t.

Leutwyler. Indeed. Wilczek. Lett. B 47. Mesons consist of a quark-antiquark pair of a particular color and its anticolor. One may wonder what would happen if we try to see a single quark with color by reaching deep inside a hadron.3 b t −0. Quarks are so tightly bound to other quarks that extracting one would require a tremendous amount of energy. or QCD.07 1 H. Rev. Each quark flavor can have three colors usually designated red. 365 (1973). 2 7 . B 667. 3 We work in natural units. green.17 mc = 1. Lett. Gell-Mann and H. where ~ is one unit of action and c is one unit of velocity. m = 4. Moreover. 30.3 MeV. so much that it would be sufficient to create more quarks. Lett. This implies that [length] = [time] = [energy]−1 = [mass]−1 . Rev. 1 (2008).1 and so the field theory has taken on the name of quantum chromodynamics. H. Gross and F. This property of quarks. Phys. and blue. M. Lett. This “charge” is a three-fold degree of freedom which has come to be known as color.e.two choices. D. Phys. spin up (↑) or spin down (↓). Phys. one with each color. antigreen. that they are always bound in groups that are colorless. it serves to distinguish them and allows the exclusion principle to hold. [Particle Data Group].5 − 3. ms = 104+26 −34 MeV. Amsler et al.5 − 6. D.. Because the color is different for each quark. the force between them becomes small. Politzer. such experiments are done at modern particle colliders and all we get is not an isolated quark. This aspect is referred to as asymptotic freedom. about 1 fm = 10−15 m) the “bare” masses of the quarks are: mu = 1. +0. 1343 (1973). Baryons are made up of three quarks.0 MeV. Phys. it was suggested that quarks possess another “charge” which enables them to interact strongly with one another. 1346 (1973). The antiquarks are colored antired. Even though quark color was originally an ad hoc idea.27+0. md = 3. and antiblue. as two quarks approach each other very closely (or equivalently have high energy).2 When probed at small distances compared to the size of a hadron (i. 30. it soon became the central feature of the theory determining the force binding quarks together in a hadron.20 GeV and m = 171. Both baryons and mesons are thus colorless or white. the color force has the interesting property that. Masses are as quoted in C. This would seem to violate the exclusion principle! Not long after the quark theory was proposed. is called confinement.07 GeV.2 ± 2.1 GeV. but more hadrons (quark-antiquark pairs or triplets). Fritzsch. J.11 −0.

with masses me = 0.67 eV at 95%CL. Rev. d.0021 GeV). the effective quark masses in composite hadrons are significantly larger. the weak and electromagnetic forces are indeed two different manifestations of a single. 22. namely. For example. 5 8 .84 ± 0. 1968) p. Weinberg. as yet unseen. The lightest flavors are generally stable and are very common in the universe as they are the constituents of protons (uud) and neutrons (ddu). More massive quarks are unstable and rapidly decay. culminating in the discovery of the predicted W ± and Z 0 bosons (mW = 80.4 One important aspect of on-going research is the attempt to find a unified basis for the different forces. the favored electroweak symmetry breaking mechanism requires the existance of a scalar Higgs boson. Suppl. 180.367. mν < 2. 8. Glashow. However. muon (µ). In these lectures.3 GeV. s. and ντ . A. However. Phys. Salam. 1. Stockholm. and tau (τ ). To understand the subtleties of our present-day view. Nucl. for u. more fundamental electroweak interaction. 330 (2009). 0. Elementary Particle Physics. mν < 0. So far as we know. They come in three flavors: electron (e). these can only be produced as quark-pairs under high energy conditions. It was Fermi who first proposed a kinematic search for the neutrino mass from the hard part of the spectra in Tritium beta decay. 0. 19.510998910 ± 0. S. Phys.000004 MeV. respectively.However. In the presence of non-vanishing leptonic mixing. and b. N. Almqvist and Wiksell. Svartholm (Nobel Symposium No.0000000013 MeV. Komatsu et al. Each flavor has an associated neutrino: νe . WMAP data provides the nominally strongest constraint on the P sum of the neutrino masses. L. Leptons are fractionally spin. we will provide an elementary introduction to quan4 E. J.3 GeV.029 GeV and mZ = 91. Astrophys. [WMAP Collaboration]. ed. νµ . 1264 (1967). we need to begin with the ideas leading up to its formulation. such as in particle accelerators and in cosmic rays. 579 (1961).658367 ± 0.9 GeV.5 GeV.17 MeV. this search sets an upper limit on the absolute mass of any of the neutrinos.403 ± 0. Nowadays physical phenomena can be discussed concisely and elegantly in terms of quantum field theories. c. mµ = 105. S. the veritable “zoo” of subatomic particles is composed of composites of quarks and leptons that interact by exchanging force carriers. 0. and mτ = 1776.5 The electroweak theory has had many notable successes.5 GeV and 4.21 particles which do not strongly interact.1876 ± 0. at present.2 eV at 95% CL. Lett.

(Wiley. The course will cover the major theoretical predictions and experimental tests. a review for more advanced theory students.1) (1. L. electroweak theory. Singapore.) In a local field theory the Lagrangian can be written as the spatial integral R of a Lagrangian density. West. Minimization condition on δS yields 0 = δS Z = d4 x [∂φ L δφ + ∂∂µ φ L δ(∂µ φ)] . The second term in the integrand can be integrated by parts.2) can be written as a surface integral over the boundary of the four dimensional spacetime region of integration.2. p. quantum chromodynamics. S = L (φ.2. D. Slansky and G. Gauge Theories of the Strong. Z 0 = d4 x[∂φ L δφ − ∂µ (∂∂µ φ L ) δφ + ∂µ (∂∂µ φ L δφ)] . 1984). 9 . 56. in The Santa Fe TASI-87 (eds. Bjorken and S.2) with δ(∂µ φ) = ∂µ (φ + δφ) − ∂µ φ = ∂µ (δφ). (McGraw-Hill. Weak and Electromagnetic Interactions. D. Quigg.6 1. (1. and is suitable as a starting point for beginning theory students. Quarks and Leptons: An Introductory Course in Modern Particle Physics. World Scientific. Phys. q˙i ) dt. S = L(qi . Relativistic Quantum Fields. New York. where the field φ itself is a function of the continuous parameters xµ . These lectures will build upon the content of many excellent textbooks. Front. 1988). Drell.3. 1 (1983).2. C. and physics of the Higgs boson. J. (We adopt the standard notation q˙i ≡ ∂t qi . is an integral of the so-called Lagrangian function from which the system’s behavior can be determined by the principle of least action. R. An Introduction to Standard Model Physics. New York. Rosner. and as an introduction to the field for experimentalists. J. 1965).tum electrodynamics (QED). Halzen and A. (1. ∂µ φ) d4x. As in the particle 6 F. Martin. The action of such a physical system. D. the last term in Eq. Using Gauss theorem.2 Canonical Quantization The state of a physical system consisting of a collection of N discrete point particles can be specified by a set of R 3N generalized coordinates qi .

2.2) leads to the Euler-Lagrange equation of motion for a field: ∂µ (∂∂µ φ L ) − ∂φ L = 0 . (1.mechanics case. the initial and final configurations are assumed given. ǫµνρσ ∂ν Fρσ = 0. ~ respectively). The Hamiltonian is defined by H= 3N X i=1 and so we can write H= Z pi q˙i − L(qi . and we have extracted the electron charge e ≡ −|e| from the four-vector current density j µ .2.2. where Aµ = (φ. for arbitrary variations δφ. we restrict our consideration to deformations δφ that also vanish on the spatial boundary of the integration region. A) ~ = −∂t A− ~ ∇φ ~ and B ~ = ∇× ~ A.7 In the interaction term.6) d3 x H (x) .2.5) by substituting the Lagrangian ~ is the four-vector potential (related to to the into (1.7) 7 We adopt Heaviside-Lorentz rationalized units. the four current is not itself a fundamental field.3). (1. in which the factors of 4π appear in 1 e2 ≈ 137 ) rather than in Maxwell’s Coulomb’s law and the fine structure constant (α = 4π equations. q˙i ) (1. Hereafter. ∂µ F µν = e j ν (1. one can obtain Maxwell equations. is called the momentum density conjugate to φ(x). π(x) = ∂φ˙ L .4) 1 LMaxwell = − Fµν F µν + eAµ j µ 4 (1. the four-current should be understood as an abbreviation of many terms expressing the electric currents of other charged fields in terms of their variables.2. (1. 10 . and so δφ is zero at the temporal beginning and end of this region.2.2. electric and magnetic fields by E µν µ ν ν µ F = ∂ A − ∂ A is the antisymmetric field strength tensor. Hence. The canonical momentum for the particle system is pi = ∂q˙i L and the corresponding quantity for a field. Eq.3) For example.

t). since these are independent fields. Annalen Phys. ∂µ φ⋆ ) (1. All commutators involving starred with unstarred fields vanish at equal times.2.g. 17. ∂µ φ. pj ] = [qi . φ⋆.11) will be assumed to hold. Einstein. e. qj ] = 0 have as their field counterparts [π(~x. ∂µ φ) . [pi . the times were set equal but not otherwise specified.8) The Heisenberg commutation relations [pi . φ⋆ (~y . t).10) and the additional commutation relation [π ⋆ (~x. A Lorentz-Poincar´e change of referencial is a real. and the field φ⋆ will have its canonically conjugate momentum.9) with all other pairs of operators commuting. linear transformation of the coordinates conserving the norm of the intervals between different points of spacetime. 891 (1905) [Annalen Phys.2. qj ] = −iδij . t)] = −iδ (3) (~x − ~y ) (1. (1. For such 8 A. t)] = −iδ (3) (~x − ~y ) . Once these are given. π ⋆ = ∂φ˙ ⋆ L . however. and therefore a change in the origin of time has no physical consequences. In the commutation relations. 11 .3 Lorentz Group One paramount prerequisite to be imposed on a theory describing the behavior of particles at high energies is that it be consistent with the special theory of relativity.2. φ(~y. 1.8 This can be achieved by demanding covariance of the equations under Lorentz-Poincar´e transformations. their values at different times are determined by the equations of motion. It is noteworthy that the commutation relations are only defined at equal times. φ(x) and φ⋆ (x). the equation ∂µ [∂∂µ φ⋆ L ]−∂φ⋆ L = 0 will too be satisfied. The Hamiltonian density will be H = π(x) φ˙ + π ⋆ (x) φ˙ ⋆ − L (φ.where ˙ H (x) = π(x) φ(x) − L (φ. If there are various classical fields to be quantized. 194 (2005)].2. (1. 14.

apparatus related to the first one by a Lorentz transformation. similarly. a measuring apparatus M. Hereafter. in QFT. similar. rotated through an angle θ about some axis and moving with some fixed velocity relative to the apparatus A. The statement of relativistic invariance is that the measurements made by M on the state |ψA i yield the same results as those made by M ′ on the state |ψA′ i. the corresponding probability is 12 . To obtain the formal consequences of this statement. which prepares the physical state |ψA′ i. for example. Consider now another. −1. where |φp i describes the state in which just this particular momentum is found for the electron.13) i.. we may ask for the probability that the electrons emitted have momentum p. we recall that in a quantum mechanical measurement we generally determine the probability that the physical system is in some state |φi. and give the name of Lorentz transformation to the homogenous transformations with aµ = 0. −1. Consider. In addition. be a black box that emits electrons through an aperture. The quantized theory will then also be Lorentz invariant if (as indeed is the case) the commutation relations transform covariantly.. aparatus A′ will be the same source.transformation.3. e. there is a transformation law for the field φ(x). satisfying x′µ x′µ = xµ xµ . The ∗ condition of reality leads to (Λµν ) = Λµν and invariance of the norm yields gµν xµ xν = gµν x′µ x′ν = gµν Λµα Λν β xα xβ . To this end. which differs from M only in that it is shifted relative to M by the same Lorentz transformation that connects A′ with A. (1. where gµν ≡ diag(1.e. Apparatus A may. it is possible to discuss Lorentz invariance in a way divorced from the specific form of the equations of motion. which is being used to make measurements on the state |ψA i and another measuring apparatus M ′ . Actually. the new spacetime coordinates x′µ are obtained from the old ones xµ according to x′µ = Λµν xν + aµ . The probability of that happening will be |hφp |ψA i|2. For the transform source and measuring apparatus. consider a system to be fixed and some apparatus that serves to prepare a physical state |ψA i. (1.3. so that transformed fields φ′ (x′ ) satisfy the same equations in the new spacetime coordinates. −1) is the metric tensor. we treat the translation of spacetime axes separately.12) gµν Λµα Λν β = gαβ .g.

. New York.3. 13 .3. the transformation is called orthochronous because it conserves the sense of timelike vectors. (1. The invariance requirement implies that |hφp′ |ψA′ i|2 = |hφp |ψA i|2 . The crucial property here is that all transformations can be expressed as a succession of infinitesimal transformations xµ → x′µ = Λµν xν = (δ µν + ω µν )xν . The proper Lorentz group is a Lie group.9 and so here we take U to be unitary. i. Gasiorowicz. where the quantities ω µν are infinitesimals and thus we only keep terms linear in ω µν . Because the measuring apparatus M and M ′ are connected by the same Lorentz transformation. Because the vector space of states contains all possible physical states. we must have both |ψA′ i = U(Λ) |ψA i and |φp′ i = U(Λ) |φp i. We thus have hψA |φ(x)|ψA i = hψA U „ (Λ)|φ(x′)|U(Λ)ψA i . if det (Λµν ) = 1. the transformation also conserves the sense of Cartesian systems in ordinary space. Inc. this will be hψA |φ(x)|ψA i.. and for the state ψA′ it will be the measurement of the expectation value of the field at the transformed point.26.3. The ensemble of these transformations forms a group dubbed proper Lorentz group.15) with x′ = Λx. If Λ00 > 0. consider the measurement of the expectation value of the scalar field φ(x). From this we can deduce that U(Λ) must be an unitary (or antiunitary) transformation.16) (arbitrarily close to the identity). For an infinitesimal transforma9 see. For a state |ψA i. Time-reversal invariance is the only symmetry requiring an antiunitary operator. Now. S. 1966) p. |ψA i and |ψA′ i must be related by some transformation U(Λ) that depends on the Lorentz transformation Λ.|hφp′ |ψA′ i|2 .14) Therefore the scalar field in a Lorentz invariant theory would transform according to φ(x′ ) = U(Λ)φ(x)U „ (Λ) (1.e. Elementary Particle Physics. for example. (John Wiley & Sons. hψA′ |φ(x′ )|ψA′ i. (1. Additionally. where |φp′ i is the state for which the electron has the momentum p′ connected to p by the same Lorentz transformation that connects A and A′ .

In the lowest-dimension non-trivial representation of the 14 .3. Now. (1. (1.19) where in the last line we have used the antisymmetry of ω µν . we obtain [Jµν .21) where ǫijk = +1(−1) if ijk are a cyclic (anticyclic) permutation of 1 2 3 and ǫijk = 0 otherwise. . θ2 . ν = 1.. called the Lie algebra of the group. (1. φ(x)] = i(xµ ∂ν − xν ∂µ ) φ(x) ≡ Lµν φ(x) . the infinitesimals are real antisymmetric tensors. and L3 = L12 are the differential operators representing the orbital angular momentum. which satisfy the commutation relations [Ji . (1. ωµν + ωνµ = 0. identifying η = 21 ω µν Jµν . If we now write U(Λ) = eiη . the transformations that lie infinitesimally close to the identity define a vector space.e.3.3.13) implies gµβ ω µσ + gσν ω νβ = 0 . (1. for an infinitesimal transformation (1. For any continuous group.3.3. the generators of the Lie algebra are the angular momentum J k . .3.15) becomes φ(x) + i[η. The basis vectors for this vector space are called generators of the Lie algebra.3. φ(x)] + · · · = φ(xµ + ω µν xν ) . For example.20) Note that for µ.17) i. θ3 ) that can be regarded as the component of a vector directed along the axis of rotation with magnitude given by the angle of rotation.18) Expanding the right hand side in terms of ω. . the condition (1. each rotation can be labeled by a set of continuosly varying parameters (θ1 . Jj ] = iǫijk Jk . 2. we obtain i[η.tion. L2 = L31 . φ(x)] ≃ φ(x) + ω µν xν ∂µ φ − φ(x) ≃ ω µν xν ∂µ φ ≃ 1 µν ω (xν ∂µ 2 − xµ ∂ν ) φ(x) . 3 the quantities L1 = L23 . where η is hermitian and reduces to zero for the identity transformation.

J ρσ ] = i(g νρ J µσ − g µρ J νσ − g νσ J µρ + g µσ J νρ ) .12 particle of spin projection up (m = 21 or ↑) and spin projection down (m = − 12 or ↓) along the 3-axis. Pauli.3) yields the Klein-Gordon equation ∂µ ∂ µ φ + m2 φ ≡ (22 + m2 )φ = 0 . that is the column vectors (10 ) and (01 ).2. 15 . The result is [J µν . (1.4 Klein-Gordon Equation The Lagrangian formulation is particularly suited to relativistic dynamics because provided our choice of L is a Lorentz scalar. substituting the Lagrangian 1 1 (1. Z. 1. For example.23) Any matrices that are to represent this algebra must obey these same commutation rules.22) 1 0 i 0 0 −1 The basis (or set of base states) for this representation is conventionally chosen to be the eigenvectors of σ3 . (1. describing a spin.25) By recalling that a prescription for obtaining Schr¨odinger equation for a free particle of mass m is to substitute the classical energy momentum relation ~ we can E = ~p 2 /2m by the differential operators E → i~∂t and p~ → −i~∇. 336 (1926).3) will be Lorentz invariant.24) L = ∂µ φ ∂ µ φ − m2 φ2 2 2 into (1. Phys. the equation of motion resulting from (1.4. We will soon see that the six J µν operators generate the three boosts and three rotations of the Lorentz group. where σi are the Pauli matrices10 ! ! ! 0 1 0 −i 1 0 σ1 = .4.3. the generators may be written Ji = 21 σi . σ3 = . To determine the commutation rules of the Lorentz algebra. 36. (1. σ2 = . 10 W.rotation group.3. respectively.2. we can now simply compute the commutators of the differential operators J µν .

4. in addition to the acceptable E > 0 solutions. Considering the motion a free particle of energy E and momentum p. Multiplying Eq.27) we find ρ = 2 E |N|2 and ~ = 2 ~p |N|2 . Acta 7. Pauli and Weisskopf gave a natural interpretation to positive and negative probability densities by inserting the charge e into (1. (1. Phys.26) Consequently.4. (1.4. and substracting. Weisskopf. and ~ is the density flux of a beam of particles.27) ~ where ρ is the probability density (|φ|2 d3 x gives the probability of finding the particle in a volume element d3 x). 16 .~x−Et) . φ = N ei(~p.11 With this in mind. (1. {z } {z } | | ρ (1.4. which 11 W. Helv. Eq.27). We note that the probability density ρ is the timelike component of a four-vector ρ ∝ E = ±(~p 2 + m2 )1/2 . described by Klein-Gordon solution. Pauli and V. We cannot simply discard the negative energy solutions as we have to work with a complete set of states.4. (1. leads the continuty equation ~ [−i(φ∗ ∇φ ~ − φ ∇φ ~ ∗ )] = 0 ∂t [i(φ∗ ∂t φ − φ ∂t φ∗ )] +∇.28) from Eq. not a probability density.25) could otherwise have been called the relativistic Schr¨odinger equation.4.4. (1. ej 0 represents a charge density.25) by −iφ∗ and the complex conjugate equation by −iφ. and so the fact that it can be negative is no longer objectable. In some sense.30) and interpreting it as the electromagnetic charge-current density. (1.29) Thus.4. we have negative energy solutions which have associated a negative probability density.see that (for ~ = 1) Klein-Gordon equation satisfies the relativistic energymomentum relation E 2 = ~p 2 + m2 . ej µ = i e (φ∗ ∂ µ φ − φ ∂ µ φ∗ ) . 709 (1934).4. (1. and this set inevitably includes the unwanted states.

4. 939 (1948). 749 (1949). equivalently.4. 14. p~). taking its antiparticle e+ of the same (E. generally referred to as the “spinless electron. St¨ uckelberg. ~p) = 2e|N|2 (−E. The prescription for handling negative energy configurations was put forward by St¨ uckelberg and by Feynman. C. 558 (1941). 74. Of course the reason why this identification can be made is simply because e−i(−E)(−t) = e−iEt . as it lies at the heart of our approach to Feynamn diagrams. P. as far as a system is concerned. 322 (1941).we will make clear in a moment. we know that the electromagnetic four vector current is ej µ (e− ) = 2e|N|2 (E. 76. −p) . we have e+ E>0 equivalent to E<0 e− time → In other words. negative-energy particle solutions going backward in time describe positive-energy antiparticle solutions going forward in time. three-momentum ~p. Feynman. G. Rev. Acta. R.12 Expressed most simply. the emission of an antiparticle with energy E is the same as the absorption of a particle of energy −E.” From (1. Consider a spin-0 particle of energy E. 12 E.32) which is exactly the same as the current of the original particle with −E. Phys. Pictorially. 17 . (1. and charge e. a positive energy antiparticle propagating forward in time.4. Phys. 23 (1942). because its charge is −e. the E < 0 solutions may then be regarded as E > 0 solutions for particles of opposite charge (antiparticles).31) Now. It is crucial to master this idea. we obtain −ej µ (e+ ) = −2e|N|2 (E. p~) . 15.28) and (1. 14. (1. the idea is that a negative energy solution describes a particle which propagates backwards in time or.30).4. Hence. Helv. −~p.

Roy. then transitions can only occur to eigenstates with the same eigenvalue. 1. We are lead to consider matrices αk (k = 1.5 Dirac Equation Let us now attempt to construct a wave equation for spin. 13 (1. A 117.34) we see that all the coefficients αi and β anticommute with each other. αl } = 2δ kl . 18 . Dirac. β} = 0. {αk . and β 2 = 1 . A. and α3 are determined by the requirement that a free particle must satisfy the relativistic energy-momentum relation (1.5. Lond. unlike the Klein-Gordon equation. and hence they cannot simply be numbers. From Eq. it must also be linear in ~ and therefore the Hamiltonian has the general form ∇. 133. 60 (1931). In order to be covariant.4. 126. is linear in ∂t . p~ + β m) ψ(x) .33) where the four coefficients β and α1 . 2. Invariance of a system under the symmetry operation g means that if the system is in an eigenstate of C.26) H 2 ψ = (αi pi + βm)(αj pj + βm)ψ = ( αi2 p2i + (αi αj + αj αi ) pi pj + (αi β + βαi ) pi m + β 2 m2 ) ψ . M. the identity I and an element g. Soc. which are required to satisfy the conditions αk αl + αl αk ≡ {αk . (1. with the condition i > j on the second term. 360 (1930).5. 351 (1928).5. Proc.21 relativistic particles of mass m. (1. 610 (1928). α2 . 118.The particle-antiparticle conjugation C constitutes a finite symmetry group containing only two elements.35) P.5. 3) and β .34) Here we sum over repeated indices. H ψ(x) = (~ α . satisfying g 2 = I. Following Dirac13 we proceed by analogy with non-relativistic quantum mechanics and write an equation which. |{z} {z } |{z} | | {z } 1 0 0 1 (1.

It is actually never necessary to have specific representation of the matrices αk and β.where 1 is the unit matrix. ∇ (1.3.5. (1.37) σ 0 0 −1I where the submatrices σ are the Pauli spin matrices (1. (1. Unless stated otherwise. i ∂t ψρ (x) = −i [αρσ ]k ∂xk ψσ (x) + m βρσ ψσ (x) .38) 0 σ 1I 0 This representation is called the Weyl or chiral representation. Hereafter we omit the spinor subscripts whenever there is no danger of confusion: ψ(x) will always stand for a column to the right of the 4 × 4 matrices. we obtain ~ ψ + mψ .40) (iγ µ ∂µ − m) ψ = 0 .36) is called a spinor. which guarantee that the relativistic energy momentum relation also holds true. The Dirac-Pauli representation is the most frequently used: ! ! 0 σ 1I 0 α ~ ≡α= and β= .39) which can be rewritten as iγ 0 ∂t ψ + iγ k ∂xk ψ − mψ = 0 . Another possible representation in a 2 × 2 block form is ! ! −σ 0 0 1I α= and β= . A four-component quantity ψα (x) which satisfies the Dirac equation. nevertheless some calculations become more transparent by the choice of a canonical form. (1. It turns out that the lowest dimensionality matrices.5. (1.5. we will always use the Dirac-Pauli representation. and ψ „ (x) for a row to the left of the matrices.5. are 4 × 4.5.22).5. Its transformation properties are different from that of a four-vector and we will study them later in this section. On multiplying Dirac’s equation by β from the left. 19 . (1. i β ∂t ψ = −i β α .41) or equivalently.

Here, we have introduced four Dirac γ-matrices, γ µ ≡ (β, βα), which satisfy
the anticommutation relations
{γ µ , γ ν } = 2g µν .

(1.5.42)

We can now unequivocally see that Dirac’s equation is actually four differential equations,
(
)
4
X
X
(1.5.43)
i [γρσ ]µ ∂µ − m δρσ ψσ = 0 ,
σ=1

µ

which couple the four components of a single Dirac spinor ψ.
As we have seen in Sec. 1.3, a general Lorentz transformation contains
rotations, and because ψ(x) is supposed to describe a field with spin, under
the transformation xµ → x′µ = Λµν xν we do not expect form invariance with
ψ ′ (x′ ) = ψ(x). We therefore allow for a rearrangement of the components.
Because both the Dirac equation and Lorentz transformation of the coordinates are themselves linear, we ask that the transformation between ψ and
ψ ′ be linear
ψ ′ (x′ ) = ψ ′ (Λx) = S(Λ) ψ(x) = S(Λ)ψ(Λ−1x′ ) ,

(1.5.44)

where S(Λ) is a 4 × 4 matrix which operates on ψ. It depends upon the
relative velocities and spatial orientations of O and O ′. S must have an
inverse, so that if O knows ψ ′ (x′ ) which O ′ uses to describe his observations
of a given physical state, he may construct his own wave function ψ(x),
ψ(x) = S −1 (Λ)ψ ′ (x′ ) = S −1 (Λ) ψ ′(Λx) .

(1.5.45)

We could equally write, using (1.5.44), ψ(x) = S(Λ−1 )ψ ′ (Λx), which provides
the identification S(Λ−1 ) = S −1 (Λ). The main problem is to find S. It must
satisfy (1.5.44) and (1.5.45). If S exists, observer O ′ , given ψ(x) by O, may
construct ψ ′ (x′ ) using (1.5.44).
By re-expressing the Dirac equation of O in terms of ψ ′ (x′ ) with the aid
of (1.5.45), O ′ could then check whether ψ ′ (x′ ) satisfies his own equation.
In other words, his equation will be form invariant provided there exists a
matrix S(Λ) such that
(iγ µ ∂µ ′ − m)ψ ′ (x′ ) = 0 ,
20

(1.5.46)

or equivalently,
(iγ µ Λµν ∂ν − m)S(Λ) ψ(x) = 0 .

(1.5.47)

(i S −1 γ µ S Λµν ∂ν − m) ψ(x) = 0 .

(1.5.48)

If we multiply by S −1 (Λ) from the left we get

The equation therefore is form-invariant, provided we can find S(Λ) such
that
(1.5.49)
S −1 (Λ) γ µ S(Λ) Λµν = γ ν .
To find S(Λ) we resort to the trick of considering an infinitesimal Lorentz
transformation. Let
i
(1.5.50)
S(Λ) = 1 − ωµν Σµν ,
2
after a bit of algebra (1.5.49) reduces to the condition
[Σµν , γ β ] = −i(g µβ γ ν − g νβ γ µ ) .

(1.5.51)

A solution is seen to be
1
i
Σµν ≡ σ µν = [γ µ , γ ν ] .
2
4

(1.5.52)

By repeated use of (1.5.42), it is easily seen that (1.5.52) satisfies the commutation relations (1.3.23) of the Lorentz algebra, i.e,
[Σµν , Σρσ ] = i(g νρ Σµσ − g µρ Σνσ − g νσ Σµρ + g µσ Σνρ ) .

(1.5.53)

Incidentally S „ (Λ) = γ 0 S −1 (Λ) γ 0 . The form for S(Λ) when Λ is not infinitesimal is
µν
S(Λ) = e−(i/2) ωµν Σ .
(1.5.54)
For a rotation ωi0 = 0 and ωij = θk , and because Σij = 21 ǫijk σ k we get
S(Λ) = e−(i/2) θ . σ , which shows the connection between ωij and the parameters characterizing the rotation (i, j, k = 1, 2, 3). For a pure Lorentz
transformation ωij = 0 and ωi0 = ϑi , and because Σ0i = 2i αi we have
S(Λ) = e(1/2) ϑ . α 
 

1 ϑ2 ϑ . α
1 ϑ2
1
+
+ ...
= 1 + ϑ.α+
2
2! 4
3! 4
2
ϑ
ϑ
= cosh + ϑˆ . α sinh .
2
2
21

(1.5.55)

For a special case, we may find the connection between ϑ and the velocity
~v characterizing the pure Lorentz transformation by looking at (1.5.49). For
example, consider a Lorentz transformation in which the new frame (prime
coordinates) moves with velocity v along the x3 axis of the original frame
(unprimed coordinates). We will leave it to the reader to convince themselves
that
t′ =
x′3

cosh(ϑ3 ) t − sinh(ϑ3 ) x3

= − sinh(ϑ3 ) t + cosh(ϑ3 ) x3

(1.5.56)

with x and y unchanged; here,
cosh(ϑ3 ) = √

1
1 − v2

and

~v
ϑˆ ≡ = kˆ .
v

(1.5.57)

Because cos(iϑ3 ) = cosh(ϑ3 ) and sin(iϑ3 ) = sinh(ϑ3 ), we see that the Lorentz
transformation may be regarded as a rotation through an imaginary angle
iϑ3 in the it-x3 plane.
To construct the currents, we duplicate the calculation of the previous
section taking into account that Dirac’s equation is a matrix equation and
thus we must consider the hermitian, rather than the complex, conjugate
equation. The Dirac’s equation hermitian conjugate is
−iγ 0 ∂t ψ „ − i∂xk ψ „ (−γ k ) − mψ „ = 0 .

(1.5.58)

To restore the covariant form we need to flip the plus sign in the second term
while leaving the first term unchanged. Since γ 0 γ k = −γ k γ 0 , this can be
accomplished by multiplying (1.5.58) from the right by γ 0 . Introducing the
adjoint (row) spinor ψ¯ ≡ ψ „ γ 0 , we obtain
¯ µ + mψ¯ = 0 .
i∂µ ψγ

(1.5.59)

Before proceeding, we pause to discuss the transformation properties of
¯ γ µ ψ(x). We have
ψ(x)
¯ S −1 (Λ) γ µ S(Λ) ψ(x)
ψ¯′ (x′ ) γ µ ψ ′ (x′ ) = ψ(x)
¯ γ α ψ(x) .
= Λµ ψ(x)
α

22

(1.5.60)

This implies that under a Lorentz transformation, the bilinear combination
¯ γ µ ψ(x) transforms like a contravariant four-vector. Along these lines,
ψ(x)
we can write down a Lagrangian describing the behavior of spin- 12 relativistic
particles of mass m
¯ µ ∂µ − m)ψ .
LDirac = ψ(iγ
(1.5.61)
Let us now resume the derivation of the continuity equation, ∂µ j µ = 0. By
multiplying (1.5.40) from the left by ψ¯ and (1.5.58) from the right by ψ, and
adding, we obtain
¯ µ ψ = ∂µ (ψγ
¯ µ ψ) = 0 ,
ψ¯ γ µ ∂µ ψ + (∂µ ψ)γ

(1.5.62)

¯ µ ψ, satisfy the conshowing that the probability and flux densities, j µ = ψγ
tinuity equation. Moreover,
¯ 0ψ = ψ„ψ =
ρ ≡ j = ψγ
0

4
X
i=1

|ψi |2

(1.5.63)

is now positive definite. In this respect, the quantity ψ(x) resembles the
Schr¨odinger wave function, and the Dirac equation may serve as a one particle
equation. In that role, however, the coefficient of −ix0 in the decomposition
Z
ψ(x) = dp ψ(p) e−ip . x
(1.5.64)
plays the role of the energy, and there is no reason why negative energies
should be excluded.
Next, we discuss the plane wave solutions of the Dirac equation. We will
treat positive and negative frequency terms separately and therefore write
ψ(x) = u(p)e−ipx + v(p) eipx .

(1.5.65)

Since ψ also satisfies
Klein Gordon equation, it is necessary that pµ pµ = m2
p
so that p0 = + p~ 2 + m2 ≡ E. Conventionally, we will call the term with
e−iEx0 the positive frequency solution. From the Dirac equation it follows
that
[iγ µ (−ipµ ) − m] u(p) e−ipx + [iγ µ (ipµ ) − m] v(p) eipx = 0
23

(1.5.66)

x . (1. (1. The two negative energy solutions (3. which we take to be         1 0 0 0  0   1   0   0          u(1) (0) =   u(2) (0) =   v (1) (0) =   v (2) (0) =   . (1. 6 b} = aµ bν {γ µ .4) u are to be associated with an antiparticle. p0 = m the equations take the form (γ 0 − 1) m u(0) = 0 (γ 0 + 1) m v(0) = 0 .70) It is easily seen that When p~ = 0. The “slash” quantities satisfy {6 a.e.5. therefore. two positive and two negative frequency solutions. Using the antiparticle prescription from the previous section: a positron of energy E and momentum p~ is described by one of the −E and −~p electron solutions.5.5. say the positron.72) . A point worth noting at this juncture.  0   0   1   0  0 0 0 24 1 (1.5.4) (−p) e−i[−p] . u(3.5.b. i. (1.69) u¯(p) (p/ − m) = 0 v¯(p) (p/ + m) = 0 . (1. x ≡ v (2. γ ν } = 2aµ bµ ≡ 2a.5.or equivalently (γ µ pµ − m) u(p) = 0 (γ µ pµ + m) v(p) = 0 .71) There are.1) (p)eip .67) because the positive and negative frequency solutions are independent..68) where p0 ≡ E > 0. The Dirac equation for a plane wave solution may thus be written as (p/ − m) u(p) = 0 (p/ + m) v(p) = 0 . It is useful to introduce the notation γ µ pµ = γµ pµ = p/. The “positron” spinors v are defined just for notational convenience.

65). we choose the covariant normalization in which we have 2E particles/unit volume. The orthogonality relations then become u¯(r) (p) u(s) (p) = m (r)„ u (p) u(s) (p) = 2mδrs E (1. where we have used (1.74) where r = 1. For fermions.5.Since (p/ + m)(p/ − m) = p2 − m2 = 0 (1. unit vol.77) 2 u¯(p) p0 u(p) − 2 m u„(p) u(p) = 0 .73) we may write the solution for arbitrary p in the form u(r) (p) = C (m + p/) u(r) (0) v (r) (p) = C ′ (m − p/) v (r) (0) .63) and (1. and C and C ′ are normalization constants.5. we obtain (1.79) u¯(r) (p) u(s) (p) = |C|2 u¯(r) (0) (m + p/)(m + p/) u(s) (0) = 2m |C|2 u¯(r) (0) (m + p/) u(s) (0) = 2m |C|2 u¯(r) (0) (m + γ 0 p0 + αk pk β) u(s) (0) = 2m |C|2 (m + E) u¯(r) (0) u(s) (0) = 2m |C|2 (m + E) δrs .75) ρ dV = ψ „ ψ dV = u„ (p) u(p) = 2E .5.5. using p/p/ = p we obtain v¯(r) (p) v (s) (p) = − (1.80) .5. v (r)„ (p) v (s) (p) = 2Eδrs .5. just as we did for bosons Z Z (1. 2. This leads to the orthogonality relations u(r)„ (p) u(s) (p) = 2Eδrs .5.76) By summing u¯(p) γ 0 (γ µ pµ − m) u(p) = 0 and u¯(p) (γ µ pµ − m) γ 0 u(p) = 0. where we have used the relation γ 0 γ k = −γ k γ 0 . E 2 Finally. 25 (1.5.5.78) and m (r)„ v (p) v (s) (p) = −2mδrs . (1.5. (1.

For E > 0 we have m + p/ χ(r) u(r) (p) = √ m+E m + σ3 E − iσ2 σ . where χ1 = (10 ) and χ2 = (01 ).~p (r) √ χ = m+E 1 = √ m+E = √ E+m m + E −σ. m+E (1. m+E (1.~p m−E ! χ(r) (E + m)−1 σ.82) Introducing two-component spinors χ(r) .5. ~p) = m + E  (1.~p σ.  ! u1 (E. ! (1. we may examine the explicit form of the solution of the Dirac equation in the Pauli-Dirac representation.~p 26    ! .~p 0 and u2 (E.81) A straightforward calculation leads to C′ = √ 1 .~p χ(r) χ(r) 0 ! .and determine the normalization constant C=√ 1 .84)  1   −1 (m + E) σ.5.5.  0  1 (1.5. ~p) = √    m+E    0 1 ! (m + E)−1 σ.5.83) and so the positive-energy four spinor solutions of Dirac’s equation are !   1   √   0 .85) .

89) = 2m 27 .86) hence the four spinor solutions of Dirac equation are u3 (E.   (1.5. We use the explicit solutions already obtained.87) and u4 (E.~p  ! m−E   0  1   . ~p) = √  1 0 !   0 1 !   −(m − E)−1 σ.5.~p m−E ! 0 χ(r) ! . For the E < 0 solutions.~p  ! m−E   1  0   . (Λ+ )αβ 2 1 X (r) (r) uα (p) u¯β (p) ≡ 2m r=1 " # X 1 = (p/ + m)u(r) (0) u ¯(r)(0)(p/ + m) 2m (m + E) r αβ   1 1 + γ0 = (m + p/) (m + p/) 2m (m + E) 2 αβ   1 1 0 m(p/ + m) + (p/ + m)[(m − p/)γ + 2E] = 2m (m + E) 2 αβ 1 (p/ + m)αβ . (1.   (1.88) To obtain the completness properties of the solutions.5. (1. ~p) = √  −(m − E)−1 σ.~p σ. the upper two components are a great deal larger than the lower ones. we consider the positive and negative solutions separately.5. 1 u(r+2) (p) = √ m+E m + E −σ.For low momenta.

5.ˆ 2 = .92) The separate matrices. but because there are four soultions.5.5. 2m (1.5. where pˆ ≡ ~p/|~p| is the unit vector pointing in the direction of momentum. The operators Λ± project positive and negative frequency solutions. To find the eigenvalues of the helicity operator we calculate14 ! ! p )2 0 1 pˆ2 0 1 (σ. 28 . (1. (1. It follows from (1.33) that the helicity operator commutes with H and therefore it shares its eigenstates with H and its eigenvalues are conserved. This projector operator h must be such that h(r) h(s) = δrs h(r) and [h(r) .ˆ p )2 0 pˆ2 14 Note that σi σj = δij + iǫijk σk .94) 2 0 σk satisfies (1. 2m r=1 (1. there must still be another projector operator. (1.~ p)2 = σi pi σj pj = (δij + ǫijk σk )pi pj = p~ 2 .90) (1. if we define Λ− by (Λ− )αβ 2 1 X (r) (r) =− v (p) v¯β (p) 2m r=1 α we get (Λ− )αβ = 1 (m − p/)αβ . Λ+ and Λ− . Λ± ] = 0 .91) The completness relation is that Λ+ + Λ− = 2 1 X (r) (r) (r) [uα (p) u¯β (p) − vα(r) (p) v¯β (p)] = 1I .5. we see that the the helicity operator.93). On inspection. and so (σ. because Λ2± = Λ± and Λ+ Λ− = Λ− Λ+ = 0. which separates the r = 1. have the properties of projection operators. Σ = pˆk .5.21 particle.Similarly.95) h = 4 4 0 (σ. we may expect the operator to be some sort of generalization of the non-relativistic operator which projects out the state polarized in a given direction for a two component spinor.5. 2 solutions. ! σk 0 1 h ≡ pˆ .5.93) Since the two solutions have something to do with the two possible polarization directions of a spin.

98) ! (E + m)−1 σ. 21 pˆ . is thus a “good” quantum number and can be used to label the solutions.5.94) simplifies to ! ! 0 1 σ3 0 1 σ3 pˆ3 = .5.~p 1 −1             ! 0 1   !   1  0 (1. the eigenvalues of the helicity operator are ( + 12 positive helicity.Thus. Assuming a particle has momentum p~ and choosing the x3 -axis along the direction of ~p.97) h= 2 2 0 σ3 p3 0 σ3 We then find hu1 = = √ √ E+m 2 E+m 2 and hu2 = = √ √ E+m 2 E+m 2  1     −1 1 −1 ! 1 0                  (E + m) −1 1 −1 0 −1 (E + m)−1 σ.99) .~p 29 1 0 ! (E + m)−1 σ. u2 . we can determine which of the four spinors u1 . −→ ⇒ h= ⇐ − 21 negative helicity.5. σ. (1. −→ (1.5.~p     !  = − 1 u2 .~p = σ3 p3 .~p    !  = 1 u1  2 1  0 σ.  2 0  −1    !   0  1 (1. σ. |~p| = p3 and the helicity operator (1. With these assumptions.96) The “spin” component in the direction of motion.5. and v2 have spin up and spin down. v1 .

5. σ. (1. In the Dirac-Pauli representation of γ 0 .49) becomes SP−1 γ 0 SP = γ 0 and SP−1 γ k SP = −γ k (for k = 1. 3).~p   0 1   !    −1 1   1 0 !  −1 (E + m)−1 σ. −1). with the positive and negative energy states (that is. 2.5.5. the ′ behavior of the four components of ψ under parity is therefore ψ1. (1.~p   0  = − 1 v2 .5.100) 2 2 0 −σ3 pˆ3 0 −σ3 We then find hv1 = = √ √  E+m     2 E+m 2 and hv2 = = √ √ E+m 2 E+m 2                   −1 (E + m)−1 σ. The “at rest” states (1. 30 . Λν µ = diag(1.~p   1 1  = v1 !  2 0  0 1 !       (1.102)  2 −1  0 −1 For space invertion.5.4 .101) 1 !   1   (E + m)−1 σ. −1.72) are thus eigenstates of parity.For antiparticles with negative energy and momentum −~p.~p  1  !   −1 0  1 1 !  0 (E + m)−1 σ.2 = ψ1. −1.~p = σ3 (−p3 ) and the helicity operator simplifies to ! ! 0 0 1 −σ3 1 −σ3 pˆ3 h= = . the electron and positron) having opposite intrinsic parities. ! (1. Then. or the parity operation. which is satisfied by SP = γ 0 .4 = −ψ3.2 and ′ ψ3.

(1. (γ 5 )2 = 1. at least two would be the same. where the 4×4 matrix is a product of γ-matrices.5. in which case the product will be reduced to three and be already included in the axial vector.103) It follows that „ γ5 = γ5.105) We are interested in the behavior of bilinear quantities under proper Lorentz transformations (that is rotations and boosts). P ¯ Scalar ψψ 1 + under P µ ¯ ψ Vector ψγ 4 Space compts.5.3: Bilinear covariants. 2 31 . + under P 5 ¯ ψ Pseudoscalar ψγ 1 − under P To construct the most general form of currents consistent with Lorentz ¯ covariance. we need to tabulate bilinear quantities of the form (Ψ)(4 × 4)(ψ).5. − under P ¯ µν ψ Tensor ψσ 6 5 µ ¯ Axial vector ψγ γ ψ 4 Space compts. To simplify the notation. No. we introduce γ 5 ≡ iγ 0 γ 1 γ 2 γ 3 . and under space invertion (the parity operation).3. Space Inversion.106) σ µν = (γ µ γ ν − γ ν γ µ ) . Because of the anticommutation relations. (1. (1. the tensor is antisymmetric i (1. (1. of Compts. The list is arranged in increasing order of the number of γ µ matrices that are sandwiched between ψ¯ and ψ.5. which have definite properties under Lorentz transformations. If five matrices were used.5.Table 1.104) In the Dirac-Pauli representation γ5 = 0 1 1 0 ! .42). An exhaustive list of the possibilities is given in Table 1. γ 5γ µ + γ µγ 5 = 0 . The pseudoscalar is the product of four matrices.

pˆ uL = −uL . (1.107) Σ0j = [γ 0 .110) (1. (1. the massless Dirac equation ! ! ~ 0 i(∂0 + σ · ∇) ψL (x) iγ µ ∂µ ψ(x) = = 0 (1. We can form two 2-dimensional representations by considering each block separately and writing ψ(ψψLR ).5. In the Weyl representation. but is the timelike component of ¯ µ ψ. Because γ 5 SP = −SP γ 5 . and so has one positive and one negative solution. The probFrom (1. 32 (1.5. In terms of ψL and ψR . changes sign under parity.5.5. Translating these results to fourcomponent form u = (uuLR ). Each solution is based on the relativistic energy-momentum relation. ~ R=0 − i(∂0 + σ . the presence of γ 5 gives rise the four vector ψγ to the pseudo-nature of the axial vector and pseudoscalar. E 2 = p~ 2 .5. with ψ(x) = u(p)e−ipx . p~ uL .109) ~ i(∂0 − σ · ∇) 0 ψR (x) divides into two decoupled equations. unlike a scalar.5.5. The two-component objects ψL and ψR are called left-handed and right-handed Weyl spinors. ~p uR . ∇)ψ 7 → EuR = σ . ∇)ψ 7 → EuL = −σ .5.112) . γ j ] = − 4 2 0 −σ j and Σjk i i = [γ j .111) for two component spinors uL(~p) and uR (~p). the boost and rotation generators can be written as ! j σ 0 i i . it is evident that the Dirac representation of the Lorentz group is reducible. γ k ] = ǫjkl 4 2 σl 0 0 σl ! . ~ L=0 − i(∂0 − σ . it follows immediately that ψψ „ ability density ρ = ψ ψ is not a scalar. a neutrino.¯ is a Lorentz scalar. The positive energy solution has E = |~p| and so satisfies σ .110) is the wave equation for a “massless” fermion. For example.60). (1. Assume (1. a pseudoscalar is a scalar under proper Lorentz transformations but.108) From the block-diagonal form of the Lorentz generators.

In the Weyl representation.5. pˆ ! u(s) . (1. ! −1 I 0 γ5 = . pˆ 0 0 σ .5.5. Symbolically. which takes νL → νR .116) γ5 σ. we can project a Dirac spinor to a left.5. (−ˆ p ) uL = uL . we consider a neutrino solution with energy −E and momentum −~p.p ~ σ. (1.5. Working in the Dirac-Pauli representation of γ-matrices.111) describes the other helicity states νR and ν¯L . 33 . and hence describes a right-handed (h = + 12 ) antineutrino of energy E and momentum ~p. These solutions break invariance under the parity operation P .110) describes νL and ν¯R . The second equation. because weak interactions do not respect parity conservation.115) u = 2 0 1I uR uR Of course. For massless neutrinos this is not a censure. the fact that 12 (1I − γ 5 ) projects out negative helicity fermions at high energy does not depend on the choice of representation. (1.p ~ χ(s) χ(s) χ(s) m+E m+E which implies γ 5 u(s) = σ . the chirality operator (γ 5 ) is equal to the helicity operator.5.83. The remaining solution has negative energy.117) where u(s) is the electron spinor of 1.113) with positive helicity.5.or right-handed spinor ! ! ! 1I 0 uL uL 1I − γ 5 u = = .5.This means that uL describes a left-handed (h = − 12 ) neutrino of energy E and momentum ~p. with E ≫ m and E ≃ |p|. p ˆ χ χ ≃ σ . we have ! ! ! (s) (s) (s) χ σ . This shows that in the extreme relativistic limit. To interpret this. (1. (1. pˆ ≃ . (1. It satisfies σ .114) 0 1I thus. we say (1. 2 0 0 uR 0 ! ! ! 0 0 uL 0 1I + γ 5 = .

(1. uL = u„L γ 0 = u„ 12 (1 − γ 5 )γ 0 = u 21 (1 + γ 5 ). 34 .121) Note that. the mixture of vector (V ) and axial vector (A) ensures that parity is violated.5. the 21 (1 − γ 5 ) in (1. u = uL + uR ≡ 21 (1 − γ 5 )u + 21 (1 + γ 5 )u . whereas the Weyl representation diagonalizes the helicity in the extreme relativistic limit (γ 5 is diagonal). because γ 5 = γ 5„ and γ 0 γ 5 = −γ 5 γ 0 . The particular advantage of the Dirac-Pauli representation is that it diagonalizes the energy in the non-relativistic limit (γ 0 is diagonal).5. (1.120) vanishes. In closing.5. yielding J µ = ue 21 (1 + γ 5 )γ µ 12 (1 − γ 5 )uν = ueLγ µ 21 (1 − γ 5 )uν . 12 (1 − γ 5 )u = uL corresponds to an electron of negative helicity. Because of the presence of the 12 (1 − γ 5 ).and so for example. we have seen no interactions and no scattering. 1. free-particle states have been eigenstates of the Hamiltonian.5. the weak current J µ = u¯e γ µ 12 (1I − γ 5 ) uν (1. (1. For example. Namely. because only left-handed neutrinos (and righthanded antineutrinos) are coupled to charge leptons by the weak interactions.119) and so (1. A vast number of experimental evidence attest that leptons enter the “charged-current” weak interactions in a special combination of two bilinear covariants. the second term in (1. We need only choose a representation if we wish to show explicit spinors. so far.120) However. In summary. since (1 + γ 5 )(1 − γ 5 ) = 0 and γ µ γ 5 = −γ 5 γ µ .5. There is no known method.5.6 Nonrelativistic Perturbation Theory Yet.118) has a V − A form. for the electron and its neutrino. parity is maximally violated.118) becomes J µ = ue 12 (1 + γ 5 )γ µ 21 (1 − γ 5 )uν + ue 12 (1 − γ 5 )γ µ 21 (1 − γ 5 )uν . (1.5.118) automatically selects a left-handed neutrino (or a right-handed antineutrino). which are discussed in Chapter 5.5. it is appropriate to peep ahead at weak interactions. Indeed.115) can be rewritten as.

For simplicity we have normalized the solution to one particle in a box of volume V . Z H0 |φn i = En |φn i with hφm |φn i = φ∗m φn d3 x = δmn . In perturbation theory we divide the Hamiltonian into two parts H0 and V (~x. that could be used to include nonlinear terms in the Hamiltonian (or Lagrangian) that will couple different Fourier modes (and the particles that occupy them) to one another. (1. t) is a perturbing interaction.125) n n or equivalently X X cn (t)V (~x. are no longer equal to the masses when interactions are present because of the possibility of self-interaction. t). where H0 is a Hamiltonian for which we know how to solve the equations of motion.122) V and V (~x. t).6. (1. we take for H0 the sum of all free particle Hamiltonians. in order to obtain a closer description of the real world. (1. n (1.6. t)φn (~x)e−iEn t = i c˙n (t)φn (~x)e−iEn t . with the physical masses appearing in them.6. Therefore. Since the only soluble field theory is the free-field theory.15 In the formal development we consider. t)]ψ = i∂t ψ (1.6. Any solution of (1.124) n n When this expression is substituted in the Schr¨odinger equation we get an equation for the coefficients cn (t) X X cn (t)V (~x.6.other than perturbation theory.6. for the sake of simplicity. 35 .123) can be expressed in the form X X |ψi = cn (t)|nie−iEn t = cn (t) φn (~x) e−iEn t .123) for such a scalar particle moving in the presence of an iteraction potential V (~x. The objective is to solve Schr¨odinger equation [H0 + V (~x. t)|nie−iEnt = i c˙n (t)|nie−iEn t . a theory involving one scalar field. inevitably we are forced to resort to some form of approximation methods.126) n 15 A point worth noting a this juncture: the quantities which in the free Lagrangians play the role of the masses of the free particles.

provided that the potential is small and transient.6. as a first approximation. (1.6. the expression for cf (t) is only a good approximation if cf (t) ≪ 1.127) we get Z (1. as this has been assumed in obtaining the result. and ci (−T /2) = 1. as desired. Tf i ≡ cf (T /2) = −i Z T /2 −T /2 dt Z  ∗ d3 x φf (~x)e−iEf t V (~x. we obtain the following coupled linear differential equations for the coefficients Z X c˙f = −i cn (t) φ∗f V φn d3 x ei(Ef −En )t . we can. assume that these initial conditions remain true at all times.Multiplying by φ∗f .129) c˙f = −i d3 x φ∗f V φi ei(Ef −Ei )t .130) cf (t) = −i dt −T /2 and.122). We therefore set at time t = −T /2 all the cn (−T /2) = 0.6. To find the amplitude for the system to be in the state |f i at t. Next.131) which can be rewritten in a covariant form as follows Z Tf i = −i d4 x φ∗ (x) V (x) φi (x) .128) n shows that the system state |ψi = |ii. at time T /2 after the interaction has ceased. 36 .132) Certainly. (1. integrating over the volume and using the orthogonality relation (1. (1.6. the particle is in an eigenstate i of the unperturbed Hamiltonian. The relation X |ψi = cn (t)|ni (1. in particular. Replacing the initial condition into (1. we project out the eigenstate |f i from |ψi by calculating Z Z t ′ ′ d3 x φ∗f V φi ei(Ef −Ei )t (1.6.6.6.127) n Assume that before the potential V acts.6. t)[φi (~x)e−iEi t ] . for n 6= i.

6.6. In particle physics. with Vf i ≡ Z (1. Z +T /2 dt ei(Ef −Ei )t −T /2 Z +T /2 −T /2 dt (1. t) = V (~x) is time independent.135) The δ-funtion in (1.6. By the uncertainty principle. we consider the case in which V (~x. (1.134) expresses the fact that the energy of the particle is conserved in the transition i → f .134) |Vf i |2 W = lim 2π δ(Ef − Ei ) T →∞ T 2 = lim 2π T →∞ |Vf i | δ(Ef − Ei ) T = 2π|Vf i|2 δ(Ef − Ei ) . (1.6. i.6.137) This equation can only be given physical meaning after integration over a set of initial and final states.134) d3 x φ∗f (~x) V (~x) φi(x) .. then using Z ∞ 1 dq eiqp = δ(p) (1.131) becomes Tf i = −iVf i Z ∞ dt ei(Ef −Ei )t −∞ = −2πi Vf i δ(Ef − Ei ) .6.6. ρ(Ef )dEf is 37 . we usually deal with situations where we begin with a specified initial state and end up in one set of final states. We denote with ρ(Ef ) the density of final states. and |Tf i |2 is therefore not a meaningful quantity. We define instead a transition probability per unit time |Tf i |2 .136) W = lim T →∞ T Squaring (1.e.It is tempting to identify |Tf i |2 with the probability that the particle is scattered from an initial state |ii to a final state |f i. To see whether this identification is possible. this means that an infinite time separates the states i and f .133) 2π −∞ (1.6.

6. order in V . The correction to Tf i is Tf i = · · · − X Vf n Vni n Z ∞ i(Ef −En )t dt e −∞ Z t ′ dt′ ei(En −Ei )t . of great practical importance. in the right-hand side of (1.143) is the perturbation series for the amplitude with terms to first. . .142) and the rate for the i → f transition is given by (1.6. Integration over this density.16 Clearly we can improve on the above approximation by inserting the result for cn (t).6. Ei − En + iǫ (1.127) " # Z t X ′ c˙f (t) = · · · + (−i)2 Vni dt′ ei(En −Ei )t Vf n ei(Ef −En )t (1. second.138) This formula.143) E − E + iǫ i n n Equation (1. (1. (Chicago: University of Chicago Press.130). . Nuclear Physics. (1. .141) dt′ ei(En −Ei −iǫ)t = i Ei − En + iǫ −∞ The second order correction to Tf i is given by Tf i = · · · − 2πi X n Vf n Vni δ(Ef − Ei ) .6. we must include a term in the exponent involving a small positive quantity ǫ which we let go to zero after integration Z t ei(En −Ei−iǫ)t ′ .6.6. 1950).6. 16 E. .139) −T /2 n where the dots represent the first order result. (1. (1.6.6.140) −∞ To make the integral over dt′ meaningful. 38 . Ef + dEf ).138) with the replacement X 1 Vf i → Vf i + Vf n Vni + . is known as Fermi’s golden rule.6. Fermi.the number of states in the energy interval (Ef . . (1. imposing energy conservation leads to the transition rate Z Wf i = 2π dEf ρ(Ef ) |Vf i |2 δ(Ef − Ei ) = 2π|Vf i |2 ρ(Ei ) .

E. Hence.1.e. ∂ µ φr ) δφr + ∂∂ µ φr L (φr . K¨ onig. 235 (1918) [arXiv:physics/0503066]. it has no explicit coordinate dependence. see e. L. ∂ µ φr ) δ(∂ µ φr )] . ∂ µ φr ). d.2) δL = [∂φr L (φr . .1. if L is translationally invariant.1 This theorem grants observed selection rules in nature to be expressed directly in terms of symmetry requirements in the Lagrangian density. . Noether. the Lagrangian changes by the amount δL = L ′ − L = ǫµ ∂ µ L . Math-phys. d.. 253 (1957). 39 . For a detailed discussion of this theorem. 1 (2. Wiss. Klasse. X (2. Phys. n. r where δφr = φr (x + ǫ) − φr (x) = ǫν ∂ ν φr (x) .3) E.1 Noether Theorem The remarkable connection between symmetries and conservation laws are summarized in Noether’s theorem: any differentiable symmetry of the action of a physical system has a corresponding conservation law.1) On the other hand. i.g. Hill. for systems described by n independent fields L (φr .1. Nachr. .Chapter 2 Symmetries and Invariants 2. (2. Gesellsch. . zu G¨ottingen. where r = 1. Under an infinitesimal displacement x′µ = xµ + ǫµ . 23. Mod.. Rev.

1.1. ǫµ ∂ µ L (φr .9) in the equations of motion and to determine whether they take the same form in the prime coordinate system as they did in the unprimed system. we can identify pν as the conserved energy-momentum four-vector.1. r From this differential conservation law one finds " # Z Z X pν = d3 x J0ν = d3 x πr ∂ν φr − g0ν L (2.3. We have already seen that J00 is the Hamiltonian density X J00 = πr φ˙ r − L = H (2. and differs from the unit matrix if the field is not a scalar field. Here. We have already seen an example of this for the Dirac equation. we can write ∂ µ Jµν = 0.16). The practical test of Lorentz invariance is to make the replacement −1 φr (x) → Srs (Λ) φs (x′ ) (2.3.10) . we may construct the angular momentum constant of motion by considering an infinitesimal Lorentz transformation.7) r and Z d3 x J00 = H . Srs (Λ) is a transformation matrix for the fields φr under the infinitesimal Lorentz transformation (1.8) Thus.5) Jµν = −gµν L + ∂∂ µ φr L ∂ν φr .50) and (1.6) r and so ∂ t pν = 0. (1.3) gives # " X (2.1. where we recall from (1.16).Equating these two expressions and using Euler-Lagrange equation (1.52) that 1 Srs (Λ) = δrs + [γ µ . ∂ µ φr ) ǫν ∂ ν φr .5. 8 40 (2. γ ν ]rs ωµν .5. Similarly.1.1.4) ∂∂ µ φr L (φr . where the energy-momentum stress tensor Jµν is defined by X (2.2. (2.1. ∂ µ φr ) = ∂ µ r Because this holds for arbitrary displacements ǫµ .

(2. ∂µ Mµνλ = ∂µ  The conserved angular momentum is Z νλ J = d3 x M0νλ Z    = d3 x xν J0λ − xλ J0ν − iπr Σνλ rs φs = i[xµ ∂ ν − ξ ν ∂ µ ]φr + Σµν rs φs (x) (2. where     νλ φs xλ g µν − xν g µλ L + ∂∂µ φr L xν ∂ λ − xλ ∂ ν φr − iσrs    (2. If we impose the requirements of Lorentz covariance on the matrix elements of 41 . 2 (2.10). we write −1 δφr (x) = Srs (Λ)φs (x′ ) − φr (x) i = φr (x′ ) − φr (x) + ωµν Σµν rs φs . (2..e. which are derived from L by an invariant action principle. ∂ µ Srs φs (x′ ) = L (φr (x′ ). leads to the energy-momentum and angular-momentum constants of motion.1.10) into the Lagrangian theory and demand that the Lagrangian density be a Lorentz scalar and hence remain form invariant under the replacement (2.13) lead to the conservation law ∂µ Mµνλ = 0. using the Euler-Lagrange equation.  −1 −1 L Ssr φs (x′ ).1. (2.1.11) about x we find.1.1. For an infinitesimal transformation.1. via the Noether theorem.14) = ∂µ xν Jµλ − xλ Jµν − i∂∂µ φr L Σνλ rs φs . L (x′ ) − L (x) = ω µν xν ∂µ L = ∂µ [∂∂µ φr L δφr ] . a scalar L guarantees Lorentz invariance of the theory and.15) so that ∂t J νλ = 0. we must ask whether we may still apply the classical result that. ∂ µ′ φr (x′ )) .1. or vectors. Going over now to the QFT.We now take over the test (2.1.12) and (2.1. in a Hilbert space. In QFT the field amplitudes φ(r) become operators upon state functions.1. i.13) Eqs.12) Expanding (2.11) This will guarantee the form invariance of the equations of motion.

Homogeneity and isotropy of spacetime imply the Lagrangian is invariant under time displacements.1. we come to certain operator restrictions on the φr (x).. however. For most field theories generally discussed in physics the Lagrangian approach and Noether’s theorem can be carried over directly to the quantum domain without difficulty.1 Maxwell-Dirac Lagrangian We have seen that a free fermion of mass m is described by a complex field ψ(x). (2. 2. is unaffected by the translation of the entire system.2 Gauge Invariance The importance of the connection between symmetry properties and the invariance of physical quantities can hardly be overemphasized.these operators.5.61) shows that ψ(x) is invariant under the global phase transformation ψ(x) → exp(iα) ψ(x). These. For a QFT a scalar L is not sufficient to guarantee relativistic invariance. transformations that commute with the spacetime components of the wave function. Inspection of Dirac’s Lagrangian (1. We have seen that these particular measurable properties of spacetime lead to the conservation of energy. To see this. In particular. are only three of various invariant symmetries in nature which are regarded as cornerstones of particle physics. 2.2.6) and (2.1.2.15) are found to be satisfactory. we will focus attention on conservation laws associated with “internal” symmetry transformations that do not mix fields with internal spacetime properties. but we must also verify that the fields obey these operator requirements. momentum.e. i.16) where the single parameter α could run continuously over real numbers. and does not change if the system is rotated an infinitesimal angle. which represent physical observables as viewed in two different Lorentz frames. the pµ and Jµν obtained through Noether’s procedure in (2. Now. and angular momentum. In this section. Noether’s theorem implies the existance of a conserved current. we need to study the invariance of L under infinitesimal U(1) 42 .

for ψ (and similarly for ψ) of an equation for a conserved current ∂µ j µ = 0. Invariance requires the Lagrangian to be unchanged. {ψa (x).2 A global phase transformation is surely not the most general invariance. for it would be more convenient to have independent phase changes at each point. δL ¯ ∂∂ ψ¯L = ∂ψ L δψ + ∂∂µ ψ L δ(∂µ ψ) + δ ψ¯ ∂ψ¯ L + δ(∂µ ψ) µ = ∂ψ L (iαψ) + ∂∂µ ψ L (iα∂µ ψ) + .2.3). which acquires an additional phase change at each point δLDirac = ψ iγ µ [i∂µ α(x)] ψ . ψb (y)} = {ψa„ (x). (2.transformations ψ → (1 + iα)ψ.61). .2. (2. that is.18) ∂∂µ ψ L ψ − ψ¯ ∂∂µ ψ¯L = ψγ 2 R using (1. (2. (2. which transforms covariantly under phase transformations.61) is not invariant under local gauge transformations.2.5.2.2. .2.2. .2.17) The term in square brackets vanishes by virtue of the Euler-Lagrange equa¯ and so (2. (2.17) reduces to the form tion.16) to include local phase transformations jµ = − ψ → ψ ′ ≡ exp[iα(x)] ψ .19) The derivative ∂µ α(x) breaks the invariance of Dirac’s Lagrangian. It follows that the charge Q = d3 x j 0 must be a conserved quantity because of the U(1) phase invariance. ψb„ (y)} = 0. where  i ¯ µψ . = 0. ψb„ (y)} = δ 3 (x − y) δab . then local gauge invariance can be restored L = ψ (i D 6 − m) ψ 6 (x) ψ . 2 43 . . = ψ (i 6 ∂ − m) ψ − e ψ A (2.21) The spinor operators ψ and ψ „ satisfy the equal-time anticommutation relations {ψa (x). (1. Dµ ψ → eiα(x) Dµ ψ.   = iα ∂ψ L − ∂µ (∂∂µ ψ L ) ψ + iα∂µ (∂∂µ ψ L ψ) + . ∂µ → Dµ ≡ ∂µ + ieAµ .20) The Lagrangian (1. but if we seek a modified derivative.5. We thus generalize Eq.

4 44 .23) which does not change the electromagnetic field strength. any change in the contribution to ΦC from the integral up to a given point x0µ will be compensated by an 3 C. Lett. in Schwinger’s early papers of QED. we must introduce a gauge field Aµ that couples to fermions of charge e in exactly the same way as the photon field. Wilson.24) for points at finite separation is known as a Wilson line. However.22) and we can ensure L = L ′ if we demand the vector potential Aµ to change by a total divergence 1 A′µ (x) = Aµ (x) − ∂µ α(x) . Rev. Bohm. 33. Yang. if ψ → ψ ′ and A → A′ .2. 445 (1974). Phys.e. This path-dependent phase was used long before Wilson’s work. Rev. (2.25) ΦC = exp −ie Aµ (x)dx the phase becomes a nontrivial function of Aµ .24) ΦAB = exp −ie Aµ (x)dx A where −eAµ (x) parametrizes the infinitesimal phase change in (xµ .4 A crucial property of the Wilson line is that it depends on the path taken and therefore ΦAB is not uniquely defined. An alternative approach to visualize the consequences of local gauge invariance is as follows.3 The integral in (2. Phys. (2.2.2. D 10. (2. a Wilson loop)   I µ .Namely. we have ′ ′ L ′ = ψ (i 6 ∂ − m) ψ ′ − e ψ A 6 ′ ψ′ = ψ (i 6 ∂ − m) ψ − ψ [6 ∂α(x)] ψ − e ψ6A′ ψ . K. by demanding local phase invariance in ψ. In other words. G. xµ +dxµ ). e (2. Fµν . Phys. (Note that for a Wilson loop. 115. if C is a closed path that returns to A (i. N. Aharonov and D. that is by construction locally gauge invariant.2.. The wave function of a particle (of charge e) as it moves in spacetime from point A to point B undergos a phase change   Z B µ . Rev. and in Y. 485 (1959). 2445 (1974).2.

[Dµ . The coefficient eAµ of the infinitesimal displacement dxµ should be replaced by a n × n matrix 5 The gauge invariance of Fµν can also be seen through the commutator of the covariant derivative. Yang and R. Phys.27) If the electromagnetic current is defined as ejµ ≡ eψγµ ψ.2. The local phase changes (2.6 Such a kind of transformation requires a generalization of local phase rotation invariance to invariance under any continuous symmetry group. N. a quark can change its color from red to blue or change its flavor from u to d). Electrodynamics is thus an Abelian gauge theory. 45 .equal and opposite contribution from the integral departing from x0µ .19) form a U(1) group of transformations.2. (2. Rev.2. Pioneer work by Yang and Mills considered that a charged particle moving along in spacetime could undergo not only phase changes.2 Yang-Mills Lagrangian If by imposing local phase invariance on Dirac’s Lagrangian we are lead to the interacting theory of QED. Dν ] = ieFµν .2. L.) To verify this claim. 191 (1954).2.5 To obtain the QED Lagrangian we need to include the kinetic term (1. 4 (2. but also changes of identity (say.5).26) where dσ µν is an element of surface area.2.2. then in an analogous way one can hope to infer the structure of other interesting theories by starting from more general fundamental symmetries. 1 L = − Fµν F µν + ψ(i 6 ∂ − m)ψ − eψ A 6 ψ. we express the closed path integral (2.4). the group is said to be Abelian.23) of Aµ (x) by a total divergence.25) as a surface integral via Stokes’ theorem I Z µ Aµ (x)dx = Fµν (x)dσ µν . 6 C. 2. Since such transformations commute with one another. which accounts for the energy and momentum of free electromagnetic fields. Mills. One can now check by inspection that the Wilson loop is invariant under changes (2. 96. this Lagrangian leads to Maxwell’s equations (1.2.

31) . to ensure that changes in phase or identity conserve probability. and consequently a path-ordering (P {}) is needed. we introuce a parameter s of the path P . and hence for infinitesimal separation between A and B we can write ΦAB = 1I + iǫ(gAaµ ta ) + O(ǫ2 ) . one may consider only transformations such that det (ΦAB ) = 1. 46 (2. the ta define a linearly independent basis set of matrices for such transformations.2. we demand ΦAB be a unitary matrix. where g is the coupling constant. (2. careful must be taken as some subtleties arise because the integral in the exponent now contains the matrices Aµ (x) which do not necessarily commute with one another at different points of spacetime. The Wilson line is then defined as the power series expansion of the exponential with the matrices in each term ordered so that higher values of s stand to the left    Z 1 dxµ Aµ (x) . i.. If we wish to separate out pure phase changes (in which Aµ (x) is a multiple of the unit matrix) from the remaining transformations. Now. Φ„AB ΦAB = 1I.−gAµ (x) ≡ −gAaµ (x)ta acting in the n-dimensional space of the particle’s degrees of freedom.2. which runs from 0 at x = A to 1 at x = B. The last condition becomes evident if we note that near the identity any unitary matrix can be expanded in terms of Hermitian generators of SU(N). (2.e. Here. Hence. the theory is said to be non-Abelian. Both the Wilson line and the Wilson loop can be generalized to YangMills transformations.30) This shows that we must consider only transformations such that a a det (eig Aµ ta ) = eig Aµ Tr(ta ) = 1. However. whereas the Aaµ are their coefficients.29) or equivalently 1I = Φ„AB ΦAB = 1I + igǫ[Aµ (x)„ − Aµ (x)] + O(ǫ2 ) .28) ΦAB = P exp ig ds ds 0 If the basis matrices ta do not commute with one another.2. (2.2.

32) where the cijk are structure constants characterizing the group.7 The SU(3) structure constants cijk = fijk are fully antisymmetric under interchange of any pair of indices √ and the non-vanishing values are permutations of f123 = 1. 125. a theory involving fermions ψ be invariant under local transformations. ψ(x) → ψ ′ (x) = V (x)ψ(x) ≡ exp [iαa (x)ta ] ψ(x) . An alternative way to introduce non-Abelian gauge fields is to demand that.33) i Fµν = ∂µ Aiν − ∂ν Aiµ + gcijk Ajµ Akν . f147 = f165 = f246 = f257 = f345 = f376 = 21 .16). A summation over the repeated suffix a is implied. but for matrices Aµ (x) that do not commute with one a another.34) or equivalently.2. Phys.25). (2. All in all. (2. corresponding to the number of independent SU(N) generators. we write the field-strength tensor.) Next. Fµν = Fµν ta .2.2. There are n2 − 1 of them. In the fundamental representation of SU(2). Rev. (2. tj ] = icijk tk . (In the fundamental representation Tr ti tj = δij /2. Aν ] . the generators are proportional to Pauli matrices (ti = σi /2). f458 = f678 = 23 . The basis matrices satisfy the commutation relations [ti . where λi /2 are the Gell-Mann matrices normalized such that Tr (λi λj ) = 2δij . Gell-Mann. by considering an infinitesimal closed-path transformation analogous to (2.35) where V is an arbitrary unitary matrix (V „ V = 1I) which we show parametrized by its general form. by analogy with (2. The generators of SU(3) in the fundamental representation are ti = λi /2. (2.2. 47 .2.2. and the structure constants are defined by the LeviCivita symbol (cijk = ǫijk ). the n × n basis matrices ta must be Hermitian and traceless.corresponding to traceless Aµ (x). we demand 7 M. for a non-abelian transformation: Fµν = ∂µ Aν − ∂ν Aν − ig[Aµ. 1067 (1962). Duplicating the preceding discussion for U(1) gauge group.

where ′ L ′ ≡ ψ (i 6 ∂ − m)ψ ′ = ψV „ (i 6 ∂ − m)V ψ = ψ(i 6 ∂ − m)ψ + iψV „ γ µ (∂µ V )ψ . The [Aµ . The field strength Fµν transforms as Fµν → F′µν = V Fµν V „ .2.36) The last term. As before. g (2.L → L ′ . as in the abelian case. Aν ] term in 48 . Namely.38) which is equal to L if we take i A′µ = V Aµ V „ − (∂µ V )V „ . under the transformation (2.2. we are ready to write down the full Yang-Mills Lagrangian for gauge fields interacting with matter fields 6 − m)ψ . (2. both sides transform as V ( )V „ under a local gauge transformation. we follow the steps of QED and add a i kinetic term. spoils the invariance of L . As in the abelian case.40) The interaction of a gauge field with fermions corresponds to a term in the interaction Lagrangian ∆L = gψ(x)γ µ Aµ (x)ψ(x). Dν ] = −igFµν .35) the Lagrangian 6 − m)ψ L = ψ(i D (2.2. To obtain propagating gauge fields. it can be compensated if we replace ∂µ by ∂µ → Dµ ≡ ∂µ − igAµ (x). L = − 12 Tr(Fµν Fµν ) + ψ(i D (2. = L + ψ[g(V „ A (2.2. it can be computed via [Dµ .37) becomes ′ L ′ ≡ ψ (i D 6 ′ − m)ψ ′ = ψV „ (i 6 ∂ + g A 6 ′ − m)V ψ 6 ′V − A 6 ) + iV „ (6 ∂V )]ψ .2.39) The covariant derivative acting on ψ transforms in the same way as ψ itself under a gauge transformation: Dµ ψ → D′µ ψ ′ = V Dµ ψ. −(1/4)Fµν F iµν to the Lagrangian.2. After reminding the reader i the representation Fµν = Fµν written for gauge group generators normalized such that Tr(ti tj ) = δij /2.

These self-interactions are a paramount property of non-Abelian gauge theories and drive the remarkable asymptotic freedom of QCD. arising solely from the kinetic term.2. Ms = 0i = 12 (↑↓ + ↓↑)    |S = 1. the composite system may have total spin S = 1 or S = 0. Each nucleon has spin 21 (with spin states ↑ and ↓). with T3 = ± 21 for protons and neutrons respectively. which leads to its becoming weaker at short distances allowing the application of perturbation theory. F. (Wiley. Ms = 1i =↑↑ q (2. yielding three.42) respectively.2. M = −1i =↓↓ s q |S = 0. Theoretical Nuclear Physics. The T = 1 and T = 0 states of the 8 J.2. 2.and four-field vertices of the form (3) ∆LK = (∂µ Aiν )gcijk Aµj Aνk (2. Each nucleon is similarly postulated to have isospin T = 12 . They have no analogue in QED and arise on account of the non-abelian character of the gauge group.2.8 Consider the description of the two-nucleon system. (2.3 Isospin Isospin arises because the nucleon may be view as having an internal degree of freedom with two allowed states.41) and (4) 2 n ∆LK = − g4 cijk cimn Aµj Aνk Am µ Aν . 1952) 49 . Blatt and V.Fµν leads to self-interactions of non-Abelian gauge fields. and so following the rules for the addition of angular momenta.43) |S = 1. New York. M. Weisskopf. the proton and the neutron. MS = 0i = 12 (↑↓ − ↓↑) . The composition of these spin triplet and spin singlet states is     |S = 1. which the nuclear interaction does not distinguish.

τ2 = . The close similarity between (2. Such a 2×2 unitary matrix is characterized by four parameters. The isospin T is the analog of the angular momentum U = eiα .46) where the three traceless hermitian canonical 2 × 2 matrices ! ! ! 0 1 0 −i 1 0 τ1 = . (2.47) are just the Pauli spin matrices. 50 . (2. T3 = 0i = 12 (ψp ψn + ψn ψp )    |T = 1.2.2. (2.2. T3 = 1i = q (1) (2) (1) (2) |T = 1.2.45) reads ψ(x) + iαi [Ti .nucleon-nucleon system can be constructed in exact analogy to spin  (1) (2)  ψp ψp   |T = 1.2.48) The rotational invariance implies that the isospin is conserved. when the common phase factor is taken out. The nucleon field operators will transform according to ! ! ! ! u u ψ (x) ψ (x) ψp (x) 11 12 p p U U −1 = ≡U .2. 9 The major difference is that in (2.45) and the way in which we would express rotational invariance9 suggests a way of characterizing the invariance. τ3 = 0 −1 i 0 1 0 (2. T . the most positively charged particle is chosen to have the maximum value of T3 .2. T3 = 0i = 12 (ψp(1) ψn(2) − ψn(1) ψp(2) ) . and a conventional way of writing a general form for U is (ommiting the phase factor) U = e(i/2)α . We will speak of an invariance under rotations in an “internal” space.45) ψn (x) u21 u22 ψn (x) ψn (x) The preservation of the commutation relations requires that the 2 × 2 matrix U be unitary.44) In general.45) the spatial coordinates are not involved. we have three parameters. τ (2. T = −1i = ψ (1) ψ (2) n n 3 q |T = 0.2. ψ(x)] = ψ(x) + 12 iαi τi ψ(x) . (2. For an infinitesimal rotation.

49) by ψ(x). . The construction of antiparticle isospin multiplets requires care.2.55) [Q.e. Therefore.57) 51 . . It is easily seen that these relations are 1 T = 2 Z d3 x ψ „ (x) τ ψ(x) . ] .50) (2.2. the charge operator for nucleons Q may be written as Z Z 3 „ 3 Q = d x ψp (x) ψp (x) = d3 x ψ „ (x) 1+τ ψ(x) . T3 = 2 Hence.51) (2..2. (2. 2 (2.56) that so that charge violates isospin conservation. (2. if we consider only protons and neutrons. ψ(x)] = τi ψ(x) 2 (2.2. are similar contributions from other fields carrying baryon number.2. Tj ] = iǫijk Tk (2. It is well illustrated by a simple example.2.52) We may introduce the baryon-number operator NB by the definition Z NB = d3 x [ψp„ (x)ψp (x) + ψn„ (x)ψn (x) + .53) where the extra terms. not written down. a rotation through π about the 2-axis.i.2. Consider a particular isospin transformation of the nucleon doublet. = e−iπ(τ2 /2) ψn ψn 1 0 ψn ψn′ (2.2. where we represent satisfied by Note that  ψp (x) ψn (x)  1 [Ti .54) Q = 12 NB + T3 . It follows from the easily derived commutation relations [Ti . Ti ] 6= 0 i = 1. Z 1 d3 x [ψp„ (x) ψp (x) − ψn„ (x) ψn (x)] . 2 (2. We obtain ! ! ! ! ! ψ ψ 0 −1 ψ ψp′ p p p = −iτ2 = .2.

For example. (2. 2.58) 1 0 ψn¯ ψn′¯ However.57) therefore gives ! ! ! 0 −1 ψp¯ ψp′¯ = .59) ψp¯ ψp′¯ 1 0 That is. We obtain ! ! ! 0 −1 −ψn¯ −ψn′¯ = .3 Higgs Mechanism In the preceding sections much importance has been attached to symmetry principles. Cψp = ψp¯. A composite system of a nucleon-antinucleon pair has isospin states   −ψp ψn¯   |T = 1. We must therefore make two changes. ψn ). T3 = 1i = q (2.2. in several areas we are still far from where we need to be.57).2.2. Applying C to (2. ψp¯) transforms exactly as the particle doublet (ψp . However. to arrange an SU(3) triplet of antiparticles so that it transforms as the particle triplet. for example. T3 = 0i = 12 (ψp ψp¯ + ψn ψn¯ ) .We define antinucleon states using the particle-antiparticle conjugation operator C. We have discussed the connection between exact symmetries and conservation laws and have found that the proviso of a local gauge invariance can serve as a dynamical principle to captain the assembly of interacting field theories. the gauge principle has lead us to theories in which all the interactions are mediated by massless bosons. T = −1i = ψ ψ 3 n p¯ q |T = 0. we want the antiparticle doublet to transform in exactly the same way as the particle doublet. the antiparticle doublet (−ψn¯ . it is not possible. First we must reorder the doublet so that the most positively chargeed particle has T3 = + 21 and then we must introduce a minus sign to keep the matrix transformation identical to (2. T3 = 0i = 12 (ψp ψp¯ − ψn ψn¯ )    |T = 1.2. Cψn = ψn¯ .60) |T = 1. This is a special property of SU(2).2. (2. while we have already mentioned 52 .

12 magnetic dipoles. in contrast to the paramagnetic case the spontaneous magnetization does not return to zero when the external field is turned off. The SO(3) symmetry is hence broken down to an axial SO(2) symmetry of rotations around the external field direction. Below the Curie temperature TC the ground state is a completely ordered configuration in which all dipoles are aligned in some arbitrary direction. The fact that the direction of the spontaneous magnetization is random and the fact that the measurable properties of the infinite ferromagnet do not depend upon its orientation are vestiges of the original SO(3) symmetry. In the absence of an external field. the situation is rather different. In these circumstances the SO(3) symmetry is said to be spontaneously broken down to SO(2). Let us first elaborate on an intuitive description of hidden symmetry. This means the SO(3) invariance is manifest. which tends to align the spins in the material. the nearest-neighbor interaction between spins (or magnetic dipole moments) is invariant under the group of spatial rotations SO(3). The spontaneous magnetization of the system is zero and there is no preferred direction in space. which exists above TC . The ground state is thus infinitely degenerate. It is thus of interest to learn how to deal with symmetries that are not manifest. In the disordered paramagnetic phase. the configuration of lowest energy have non-zero spontaneous magnetization. In the infinite ferromagnet mentioned above.that the carriers of the weak force are massive. However. A particular direction for the spontaneous magnetization may be chosen by imposing an external field which breaks the SO(3) symmetry explicitely. when the system is in the ordered ferromagnetic phase. an infinite crystalline array of spin. The full symmetry is restored when the external field is turned off. it is crucial that 53 . For the rotational invariance to be broken spontaneously. the medium displays an exact symmetry in the absence of an external field. belying the rotation invariance of the underlying interaction. there are many situations in physics in which the exact symmetry of an interaction is hidden by the circumstances. because the nearest-neighbor force favors the parallel alignment of spins. A priviliged direction may be selected by imposing an external magnetic field. perhaps in the hope of evading the conclusion that interactions must be mediated by massless gauge bosons. Moreover. For temperatures below TC . The canonical example is that of a Heisenberg ferromagnet.

the potential has a unique minimum at φ = 0 that corresponds to the ground state. If the potential is an even functional of the scalar field.10 To deeply fathom the situation.63) 2 The state of lowest energy is thus seen to be one for which the value of the field φ is constant.3. Two qualitatively different cases.. To explore possibilities.) For µ > 0. it is convenient to consider an explicit potential.3. (We usually refer to hφi0 as the vacuum expectation value of the field φ.the ferromagnet be infinite in extent.61) into (1. V (φ) = V (−φ). 508 (1964). Rev. Phys. i. This identification is perhaps most easily seen in the Hamiltonian formalism. Higgs. may be distingusihed depending on the sign of the coefficient µ2 .62) where λ is required to be positive so that the energy is bounded from below.2.3. the vacuum. so that rotation from one degenerate ground state to another would require the impossible task of rotating an infinite number of elementary dipoles. W. spontaneous symmetry breaking can arise when the Lagrangian of a system possesses symmetries which do not however hold for the ground state of the system. The Higgs mechanism is a gauge theoretic realization of such spontaneous symmetry breaking. corresponding to manifest or spontaneously broken symmetry. let us now consider a simple world consisting just of scalar particles described by the Lagrangian L = 1 2 (∂µ φ) (∂ µ φ) − V (φ) (2.8) we obtain   2  1 2 ~ H = (∂0 φ) + ∇φ + V (φ) . the vacuum obeys 10 P. All in all.61) is invariant under the symmetry operation which replaces φ by −φ.3. which we denote by hφi0 . Lett. Substituting (2. it corresponds to the absolute minimum (or minima) of the potential V (φ).e.61) and study how the particle spectrum depends upon the effective potential V (φ). The value of this constant is determined by the dynamics of the theory. If µ2 > 0.3. 13. 54 . (2. V (φ) = 21 µ2 φ2 + 14 λφ4 . (2. then the Lagrangian (2.

62)] and L ′ of (2.3. we 55 . If we could solve the two Lagrangians exactly.64) The φ4 term shows that the field is self-interacting. with hφi0 = 0. because the four-particle vertex exists with coupling λ. Indeed. we will see that whatever is our choice the symmetry of the theory is spontaneously broken: the parity transformation φ → −φ is an invariant of the Lagrangian.65) cannot change the physics. Because of the parity invariance of the Lagrangian.3.65) into (2. Perturbative calculations should involve expansions around the classical minimum.) The potential has two degenerate lowest energy states. the relative sign of the η 2 term and the kinetic energy is √ negative.65) where η(x) represents the quantum fluctuations about this minimum. but not of the vacuum state. Using L . To this end. the ensuing physical consequences must be independent of this choice. with v = −µ2 /λ.24) gives mη = 2λv 2 = −2µ2 .66) are completely equivalent. If µ2 < 0. Before proceeding. Substituting (2.p Identifying the first two ′ terms of L with (1. These 2 2 minima p satisfy φ(µ + λφ ) = 0. we do perturbation theory and calculate the fluctuations around the minimum energy.3.the reflection symmetry of the Lagrangian.61) [with (2.3.3. Instead.4. the Lagrangian has a mass term of the wrong sign for the field φ and the potential has two minima. The approximate form of the Lagrangian appropriate to study small oscillations around this minimum is that of a free particle with mass µ. we pause to stress that the Lagrangian L of (2.61). we therefore write φ(x) = v + η(x) . (2. we obtain L ′ = 12 (∂µ η)(∂ µ η) − λv 2 η 2 − λvη 3 − 41 λη 4 + const.3. (2.3.3. L = 12 [(∂µ φ)(∂ µ φ) − µ2 φ2 ] . they must yield identical physics. and are therefore at hφi0 = ±v. in QFT we are not able to perform such a calculation. The higher-order terms in η represent the interaction of the η field with itself. However. either of which may be chosen to be the vacuum. (2.3. A transformation of the type (2. (The extremum φ = 0 does not correspond to the energy minimum. let us choose hφi0 = +v. Nevertheless.66) The field η has a mass term of the correct sign.

Intuitively. L ′ provides the correct physical framework.70) The third term has the form of a mass term (− 12 m2η η 2 ) for the η field. the theory contains a massless scalar. it is easily seen 56 . in attempting to generate a massive gauge boson we have encountered a problem: the spontaneously broken gauge theory seems to be plagued with its own massless scalar particle. whereas L does not. with Lagrangian density L = (∂µ φ)∗ (∂ µ φ) − µ2 φ∗ φ − λ(φ∗ φ)2 .3. there is a circle of minima of the potential V (φ) in the φ1 -φ2 plane of radius v such that φ21 + φ22 = v 2 with v 2 = −µ2 /λ.3.3. so-called “Goldstone boson. (2. By considering λ > 0 and µ2 < 0. In other words. Therefore.3. To approach our destination of generating a mass for the gauge bosons.67) which is invariant under the transformation φ → eiα φ. As before.67) as L = 1 2 (∂µ φ1 ) (∂ µ φ1 )+ 21 (∂µ φ2 ) (∂ µ φ2 )− 12 µ2 (φ21 +φ22)− 14 λ(φ21 +φ22 )2 . ξ 3 ) + O(η 4 . we translate the field φ to a minimum energy position. ξ 4 ) . we duplicate the above procedure for a complex scalar field. which without loss of generality we may take as the point φ1 = v and φ2 = 0.67) and obtain L ′ = 21 (∂µ ξ)2 + 12 (∂µ η)2 + µ2 η 2 + const. φ = √12 (φ1 +iφ2 ).find out that the perturbation series does not converge because we are trying to expand about the unstable point φ = 0. (2. the scalar particle (described by the in-principle-equivalent Lagrangians L and L ′ ) is massive.68) In this case. the η-mass is again mη = −2µ2 . (2.” Therefore. In perturbation theory. We expand L around the vacuum in terms of fields η and ξ by substituting q φ(x) = 12 [v + η(x) + iξ(x)] (2.69) into (2. we rewrite (2. The first term in L ′ stands for the kinetic energy of ξ. Therep fore. That is L possesses a U(1) global gauge symmetry.3. + O(η 3 . The correct way to proceed is to adopt L ′ and expand in η around the stable vacuum hφi0 = +v. but there is no corresponding mass term for the ξ field.3.

3. 154 (1961). for the case µ2 < 0 and λ > 0. the reason for its presence. implying a massless mode. (aside from the φ4 self-interaction term) this is just 11 J. The Lagrangian (2. Goldstone. which states that in a spontaneous symmetry breaking the original symmetry is still present.1. Nuovo Cim.70) is a simple example of the Goldstone theorem.1: The potential V (φ) for a complex scalar field. but nature manages to camouflage the symmetry in such a way that its presence can be viewed only indirectly. 57 . J. The gauge invariant Lagrangian is thus L = (∂ µ − ieAµ )φ∗ (∂µ + ieAµ )φ − µ2 φ∗ φ − λ(φ∗ φ)2 − 14 Fµν F µν . As shown in Fig.11 In the ferromagnet example. there is no resistance to excitations along the ξ-direction. we must start with a Lagrangian that is invariant under a local U(1) transformation φ(x) → eiα(x) φ(x).3. the potential in the tangent ξ direction is flat. Phys. Rev. A.71) As usual there are two cases. depending upon the parameters of the effective potential. 19. Weinberg. Goldstone. 127. To this end. (2. If µ2 > 0. Salam and S. 965 (1962). This is accomplished by replacing ∂µ by a covariant derivative Dµ = ∂µ + ieAµ .V () 2 1   Figure 2. the analogue of our Goldstone boson is the long-range spin waves which are oscillations of the spin alignment. 2. where the gauge field transforms as Aµ (x) → Aµ (x) − ∂µ α(x)/e. The final step of this section is to study spontaneous symmetry breaking of a local U(1) gauge symmetry.

3.72) we have mξ = 0. However. The Lagrangian 58 . and mA = ev. Namely. On substituting (2.72). mη = 2λv 2 . and √ a massive vector Aµ . the apparent extra degree of freedom is actually spurious.3. because it corresponds only to the freedom to make a gauge transformation.3. we have L” = + 1 (∂ H)2 − λv 2 H 2 + 21 e2 v 2 A2µ − λvH 3 2 µ 1 2 2 e Aµ H 2 + ve2 A2µ H − 41 Fµν F µν .73) φ ≃ 12 (v + η) eiξ/v . we have clearly raised the polarization degrees of freedom from 2 to 3. but we still are facing the problem of the ocurrence of the Goldstone boson.71) becomes L′ = 1 (∂ ξ)2 2 µ 2 + 12 (∂µ η)2 − v 2 λη 2 + 12 e2 v 2 Aµ Aµ + evAµ ∂ µ ξ − 41 Fµν + interaction terms . This implies that we have generated a mass for the gauge field. we first note that to lowest order in ξ (2. In other words. Actually.69) can be rewritten as q (2. This suggests that we should substitute a different set of real fields H. because it can now have a longitudinal polarization. By giving mass to Aµ . a massive scalar η. from (2. Aµ . This is a particular choice of gauge.3. The situation when µ2 < 0 is that of spontaneously broken symmetry and demands a closer analysis.3. 2 − 41 λH 4 (2.72) The particle spectrum of L ′ appears to be a massless Goldstone boson ξ. where q 1 ∂µ θ φ → 12 [v + H(x)]eiθ(x)/v .74) The Goldstone bosson is not actually present in the theory.the QED Lagrangian for a charged scalar particle of mass µ. θ. with θ(x) chosen that H is real.71). (2. To find a gauge transformation which eliminates a field from the Lagrangian. We deduce that the fields in L ′ do not all correspond to distinct particles.3. Aµ → Aµ − ev into the original Lagrangian (2. the particle spectrum we assigned before to L ′ must be incorrect. We therefore anticipate that the theory will be independent of θ.3. because of the presence of a term off-diagonal in the fields Aµ ∂ µ ξ. the Lagrangian (2.3. Actually.69). this time care must be taken in interpreting the Lagrangian (2.3.

(2. where the sub-indeces L and R indicate the fermion chirality. 2)1/6 . a left-handed lepton field LL is a singlet of the SU(3) color group. 1)2/3 . 1)0 . The gauge interactions are mediated by eight SU(3) color gluons Gaµ (8.4. 2)−1/2 . 3. and g ′ . ER (1. The unwanted massless Goldstone boson has been turned into the longitudinal polarization of Aµ . electromagnetic. A single generation of quarks and leptons consists of five different representations of the gauge group QL (3. (2. and the strength of the interactions are described by their coupling constants gs . 3)0 . and carries hypercharge −1/2 under the U(1) group. Higgs) and tiL = (τ i /2.118) and anticipating a possible SU(2) structure for the weak currents. for example. and one U(1) hypercharge gauge field Bµ (1. three SU(2) left chiral gauge bosons Aiµ (1.4 Standard Model of Particle Physics The standard model of weak. The gauge interactions arise through the covariant derivative " Dµ = ∂µ − i gs 8 X Gaµ tC a +g a=1 3 X i=1 # Aiµ tLi + 12 g ′Bµ . a doublet of the SU(2) weak isospin. 59 with i = 1.75) where taC = (λa /2. DR (3. φ(1. This is known as the “Higgs mechanism.describes just two interacting massive particles.76) .4. singlets). 0) for (quarks.5. 1)0 .” 2. The model contains a single higgs boson doublet. 0) for SU(2) (doublets. and strong interactions is based on the gauge group SU(3)C × SU(2)L × U(1)Y . g. with three fermion generations. 2)1/2 . whose vacuum expectation value breaks the gauge symmetry into SU(3)C × U(1)EM . All the above gauge bosons are realized in the adjoint representations of their corresponding gauge groups. usually referred to as a Higgs particle. UR (3. a vector gauge boson Aµ and a massive scalar H. 1)−1/3 . LL (1. we are led to construct an “isospin” triplet of “weak currents” Jµi (x) = 1 2 u¯L γµ τi uL . Proceeding in the same spirit of (1. 1)−1 . Our notation here means that. 2. We first focus attention on the electroweak sector. lepton.

e.4..2. Without loss of generality. is in the spinor representation of the SU(2)L group and has a charge 1/2 under U(1)Y .for the spinor operators.79) We repeat the by now familiar procedure of translating the field φ to a true ground state. uL = QL = uL dL ! . such that v 2 = −µ2 /λ and φ„ φ = |φ+ |2 +|φ0|2 . if you recall.80) The values of (Re φ+ . We can now expand φ(x) about this particular vacuum ! 0 1 .78) = 12 φ= 0 φ3 + iφ4 φ The gauge invariant Lagrangian is thus Lφ = (∂µ φ)„ (∂ µ φ) − µ2 φ„ φ − λ(φ„ φ)2 . we entertain the mechanism of spontaneous symmetry breaking. i. φ1 = φ2 = φ4 = 0. Im φ0 ) can range over the surface of a 4dimensional sphere of radius v. µ2 ∂Lφ 1 2 2 2 2 „ = 0 ⇒ φ (φ + φ + φ + φ ) = − . Im φ+ . i. uL = LL = νeL e− L ! . φ ≡ 2 3 4 ∂(φ„ φ) 2 1 2λ (2. i. Consider the complex scalar Higgs boson field. to approach the goal of generating a mass for the gauge bosons. (2. φ23 = −µ2 /λ.4.e. Therefore.77) R whose corresponding charges T i = J0i (x) d3 x generate an SU(2)L algebra.4. see (2. a group SO(4) isomorphic to SU(2) × U(1).81) hφi = √ 2 v 60 .e.. which. ! ! q φ + iφ φ+ 1 2 .4. The presence of mass terms for Aiµ destroy the gauge invariance of the Lagrangian.55). we define the vacuum expectation value of φ to be a real parameter in the φ0 direction. We must expand φ(x) about a particular minimum.4. Re φ0 . The vacuum expectation value is obtained by looking at the stationary points of Lφ . This implies that Lagrangian of φ is invariant under rotations of this 4dimensional sphere. (2. (2.. (2.

4.To introduce electroweak interactions with φ. there are three massive vector bosons: 1 Wµ± = √ (A1µ ∓ iA2µ ) 2 and Zµ0 = p 1 g2 + g ′2 (gA3µ − g ′ Bµ ) . 8 (2. (2.75) in the Lagrangian (2.83) = 2 z µ 2 Therefore. The relevant terms are: !     0 1 1 ′ 1 ′ µ 1 j 1 kµ ∆L = (0 v) gA τj + g Bµ gA τk + g B 2 2 µ 2 2 2 v 2  ! gA3µ + g ′ Bµ g(A1µ − iA2µ ) 0 1   = (0 v)   8 v 1 2 3 ′ g(Aµ + iAµ ) −gAµ + g Bµ = 1 2 2 1 2 v [g (Aµ ) + g 2 (A2µ )2 + (−gA3µ + g ′ Bµ )2 ] .82) Note that 1 2 2 3 2 1 2 v [g (Aµ ) − 2gg ′A3µ B µ + g ′2 Bµ2 ] = v [gA3µ − g ′Bµ ]2 + 0[g ′A3µ + gBµ ]2 8 8 1 2 2 1 m Z + mA A2µ . (2. We identify this field with the electromagnetic vector potential. The gauge fields have “eaten up” the Goldstone bosons and become massive. orthogonal to Zµ0 .4.84) (2.4.4. (2. and evaluate the resulting kinetic term (Dµ φ)„ (D µ φ).86) remains massless. Aµ = p 1 g 2 + g ′2 (g ′A3µ + gBµ ) . we replace ∂µ by the covariant derivative (2.4.79). The scalar degrees of freedom become the longitudinal polarizations of the massive vector bosons. at the vacuum expectation value hφi.4. The spontaneous symmetry breaking rotates 61 .85) The fourth vector field.4.

the four SU(2)L × U(1)Y gauge bosons to their mass eigenstates by means of the gauge interaction term of the Higgs fields. Bµ } → Aµ .4.4. we can use the previous formulae to rewrite the covariant derivative (2. {A1µ .4. (2. cos θw (2. A2µ } → {Wµ+ . θw (defined by tan θw = g ′/g). can be rewritten as mW = g gv = √ mH 2 2 2λ and mZ = mW .91) 62 .90) g 2 + g ′2 with the electron charge. After we identify the electric charge quantum number in (2.89) − ip 2 ′2 g +g where T ± = T 1 ± iT 2 .88) showing that the Higgs mass mH sets the electroweak mass scale.4. (2.4. it becomes evident that the electromagnetic interaction (a U(1) gauge symmetry with coupling e) “sits across” weak isospin (an SU(2)L symmetry with coupling g and weak hypercharge (a U(1) symmetry with coupling g ′).4. Wµ− } and {A3µ . Note that the two couplings g and g ′ can be replaced by e and θw . After identifying the coefficient of the electromagnetic interaction gg ′ e= p = g sin θw . (2.75) becomes g 1 Dµ = ∂µ − i √ (Wµ+ T + + Wµ− T − ) − i p Zµ (g 2 T 3 − g ′2 Y ) 2 ′2 2 g +g ′ gg Aµ (T 3 + Y ).4. In terms of the mass eigenstates the covariant derivative (2.4.89) with Q = T 3 + Y . Zµ0 .89) in the form g g Zµ (T 3 − sin2 θw Q) − ieAµ Q . In terms of the weak mixing angle. where the parameter θw is to be determined by the experiment. at lowest order in perturbation theory. Dµ = ∂µ − i √ (Wµ+ T + + Wµ− T − ) − i cos θw 2 (2. the photon and Z 0 -boson fields read ! ! ! Zµ0 cos θw − sin θw A3µ = .87) Aµ sin θw cos θw Bµ The W ± and the Z 0 boson masses.

the assignments Y = −1/2 and Y = +1/6.4.4.1: Weak isospin. (2. For the left-handed fields.92) as L ¯ L (i6 ∂)LL + E¯R (i6 ∂)ER + Q ¯ L (i6 ∂)QL + U¯R (i6 ∂)UR + D ¯ R (i6 ∂)DR = L +µ −µ + g(Wµ+ JW + Wµ− JW + Zµ0 JZµ ) + eAµ j µ .4. Lepton T T3 Q Y Quark T T3 Q Y 1 1 1 1 2 1 νe 0 − 12 uL 2 2 2 2 3 6 1 1 1 1 1 1 1 − −1 − − − d e− L L 2 2 2 2 2 3 6 2 2 uR 0 0 3 3 1 1 e− d 0 0 − 0 0 −1 −1 − R R 3 3 The covariant derivative (2.4.91) we can rewrite (2.93) . and hypercharge quantum numbers. The weak isospin and hypercharge quantum numbers of leptons and quarks are given in Table 2. JW = √ (¯ 2 1 −µ eL γ µ νL + d¯L γ µ uL ). For the right-handed fields.92) To work out the physical consequences of the fermion-vector boson couplings. where 1 +µ νL γ µ eL + u¯L γ µ dL ).4. T 3 = 0 and hence Y = Q. the Lagrangian for the weak interactions of quarks and leptons follows directly from the charge assignments given above. The fermion kinetic energy terms are ¯ L (i D ¯ L (i D ¯ R (i D L =L 6 )LL + E¯R (i D 6 )ER + Q 6 )QL + U¯R (i D 6 )UR + D 6 )DR . If we ignore fermion masses.92) in terms of the vector boson mass eigenstates.1. JW = √ (¯ 2 63 (2. combine with T 3 = ±1/2 to give the correct electric charge assignments. Using the form of the covariant deivative (2. we should write (2.91) uniquely determines the coupling of the W ± and Z 0 fields to fermions. respectively.Table 2. LL and QL . once the quantum numbers of the fermion fields are specified.4.

Imposing the gauge symmetry has required that the kinetic energy term in LQCD is not purely kinetic but includes an induced self-interaction between the gauge bosons.2. The remaining two terms show the presence of three.94) 3 3 and equivalent expressions hold for the other two generations.95) where q1 .4. follows simply from demanding that the Lagrangian be invariant under local phase transformations to the quark fields. 64 .and four-gluon vertices in QCD and reflect the fact that gluons themselves carry color charge.4. it is not surprising that eight vector gluon fields are needed to compensate all possible phase changes.JZµ jµ       1 1 µ 2 µ νL + e¯L γ − + sin θw eL + e¯R γ µ sin2 θw eR = ν¯L γ 2 2     1 2 2 2 2 µ µ + u¯L γ − sin θw uL + u¯R γ − sin θw uR 2 3 3      1 2 1 1 1 2 µ µ ¯ ¯ sin θw dR . Just as for the photon. + dL γ − + sin θw dL + dR γ 2 3 3 cos θw     2 1 µ µ µ ¯ = e¯ γ (−1)e + u¯ γ + u+dγ − d (2. 2. q2 . They describe the free propagation of quarks and gluons and the quark-gluon interaction. Using (2. They have no analogue in QED and arise on account of the non-Abelian character of the gauge group. Because we can arbitrarily vary the phase of the three quark color fields. Gν ] term.96) The first three terms have QED analogues. (2.2.2. local invariance requires the gluons to be massless. the field strength tensor Gaµν has a remarkable new property on account of the [Gµ . for simplicity.40) we obtain LQCD = q¯j (iγ µ ∂µ − m)qj + gs (¯ qj γ µ ta qj )Gaµ − 14 Gaµν Gµν a . (2. and q3 denote the three color fields and.4. The gauge invariant QCD Lagrangian for interacting colored quarks q and vector gluons Gµ .95) in the symbolic form LQCD = ”¯ q q” + ”G2 ” + gs ”¯ q qG” + gs ”G3 ” + gs2 ”G4” . As we anticipated in Sec. with coupling specified by gs .4. we show just one quark flavor. This becomes clear if we rewrite Lagrangian (2.

We spontaneously break the symmetry and substitute ! q 0 φ = 12 (2. Phys. after spontaneous symmetry breaking has taken place.4.99) We choose Ye so that Ye v me = √ 2 and thus generate the required electron mass. Jap. The other three fields can be gauged away. 48 (1935). LeYukawa = −Ye (¯ where Ye is the Yukawa coupling constant of the electron. The quark masses are generated in the same way. v (2. we 12 H. 17.4. we include the following SU(2) × U(1) gauge invariant term in the Lagrangian i h  + (2. e¯)L φφ0 eR + e¯R (φ− .12 For example. that because Ye is arbitrary.78). Soc.101) Note however. Proc. to generate the electron mass.97).4. 65 . 2 2 (2. On substitution of φ. The only novel feature is that to generate a mass for the upper member of the quark doublet.100) (2. the masses of the chiral fields arise from the Yukawa interactions which couple a right-handed fermion with its left handed doublet and the Higgs field after spontaneous symmetry breaking. the Lagrangian contains an interaction term coupling the Higgs scalar to the electron. Yukawa.98) v + H(x) into (2.Since explicit fermion mass terms violate the gauge symmetries. the actual mass of the electron is not predicted.97) νe . (2. the Lagrangian becomes Ye Ye eL eR + e¯R eL ) − √ (¯ LeYukawa = − √ v(¯ eL eR + e¯R eL )H .4.4. LeYukawa = −me e¯e − me e¯eH . φ¯0 ) (νee )L .4. Math. Besides the mass term.4. The Higgs doublet has exactly the required SU(2) × U(1) quantum numbers to couple to e¯L eR . The neutral Higgs field H(x) is the only remnant of the Higgs doublet.

where U(1)B is the baryon number symmetry and U(1)e.must construct the complex conjugate of the Higgs doublet q  ¯0   −φ 2 ∗ 1 v+H ¯ .103) All in all. we will describe a number of the most important theoretical results.104) where ij are the generation indices. It is important to note that the standard model also comprises an accidental global symmetry U(1)B × U(1)e × U(1)µ × U(1)τ . both offspring of QED. (2.c. it can be used to construct a gauge invariant contribution to the Lagrangian     ¯ L φ+ dR + Yu (¯ ¯ L −φ¯¯−0 uR + h. Of course. LqYukawa = −Yd (¯ u.4. It is a consequence of the gauge symmetries and the low energy particle content.τ are three lepton flavor symmetries. It is possible (but not necessary).c. To convey an impression of how the theories developed. d) φ0 φ ¯ − mu u¯u − = −md dd md ¯ ddH v − mu u¯uH v . with total lepton number given by L = Le + Lµ + Lτ . 66 . and carry on to QCD and the electroweak theory. We will start from the most precisely tested theory in physics. It is an accidental symmetry because we do not impose it. .102) breaking Because of the special properties of SU(2). namely. (2. that effective interaction operators induced by the high energy content of the underlying theory may violate sectors of the global symmetry. and how the standard model has successfully confronted experiment. φ¯ transforms identically to φ (but has opposite weak hypercharge to φ. the Yukawa Lagrangian then takes the form −L Yukawa = Yd ij QL i φ DRj +Yu ij QL i φ¯ URj +Ye ij LL i φ ERj +h. Therefore. φ = iσ φ = φ− |{z} −→ 2 0 (2. however. Up to now we have “concocted” a standard model of particle physics from general group-theory considerations of principles of symmetry and invariants.µ. Y = −1/2).4. QED. d) u . in real life a model of nature is usually uncovered in a less pristine fashion.4.

charged particles. 3. and including an interaction Hamiltonian to describe how the particles deflect one another. 3.1. we have introduced 67 . When calculating scattering cross sections. and the various virtual contributions containing one-loop and two-loops (with a closed electron loop) are shown in Figs. for many years it was difficult to conceive how such action-at-a-distance between charges came about. These virtual photons.g. The quantum field theory approach visualizes the force between electrons as an interaction arising in the exchange of “virtual” photons. we have such a tangible connection. which can only travel a distance allowed by the uncertainty principle. the lowest order (tree level) diagrams for Bhabha scattering (e+ e− → e+ e− ) are shown in Fig. The amplitude for scattering is the sum of each possible interaction history over all possible intermediate particle states.2 and 3. The number of times the interaction Hamiltonian acts is the order of the perturbation expansion.Chapter 3 QED 3. the interaction between particles can be described by starting from a free field which describes the incoming and outgoing particles.3. In QFT. However. In the non-relativistic limit of perturbation theory. e.. of course. The perturbative series can be written as a sum over Feynman diagrams. such as electrons. cannot live an existence independent of the charges that emit or absorb them. interact through their electromagnetic fields.1 Invariant Amplitude In Maxwell’s theory of electromagnetism.

(1. including their antiparticles. [For details. a factor like Vni for each interaction vertex and for the propagation of each intermediate state we have introduced a “propagator” factor like 1/(Ei −En ).6.] The intermediate states are virtual in the sense that the energy is not conserved. Figure 3. we refer the reader to Eq. Throughout this chapter we generalize the perturbative scheme to handle relativistic particles. but there is energy conservation between the initial and final states as indicated by the delta function δ(Ef − Ei ).2: Bhabha scattering one-loop diagrams. 68 . En 6= Ei .1: Bhabha scattering tree-level diagrams.143).e− e− pC pA pB e+ e+ pD e− e− pA pC pB pD + e+ e t-channel s-channel Figure 3.

3: Bhabha scattering two-loop diagrams with a closed electron loop. 69 .Figure 3.

Spinless hadrons exist (e. which acts on both Aµ and φi .g. can be 70 . The amplitude for the scattering of e− from a state φi to φf of an electromagnetic potential Aµ is Z Tf i = −i φ∗f (x) V (x) φi (x) d4 x Z = −i φ∗f ie(Aµ ∂µ + ∂µ Aµ ) φi d4 x . Working to lowest order. by choosing the interacting particles to be “spinless” charged leptons. the motion of a particle ~ is obtained by the of charge e in an electromagnetic potential Aµ = (φ. In classical electrodynamics. The corresponding quantum mechanical substitution is therefore i∂ µ → i∂ µ − eAµ . No spinless quark or lepton has ever been observed in an experiment. The spin-0 leptons (that is leptons satisfying the Klein-Gordon equation) are completely fictitious objects.4.3) The derivative in the second term. A) µ µ µ substitution p → p − eA .2) where is the (electromagnetic) perturbation.12 quarks and spin-1 gluons.2). We first have to figure out how to use perturbation theory in a covariant way.1. We illustrate this. we neglect the e2 A2 term in (3.1) V = ie(∂µ Aµ + Aµ ∂µ ) − e2 A2 (3. The Klein-Gordon equation becomes (∂µ ∂ µ + m2 )φ = −V φ (3. shown in Fig.1. This choice requires some explanation. the π-meson).φf φi Aµ Figure 3.. 3. but they are complicated composite structures of spin.1. Consider the scattering of a spinless electron in an electromagnetic potential. as it is desirable to begin by avoiding the complications of their spin.1. (3.4: A “spinless” electron interacting with Aµ .

The calculation is a straightforward extension of the previous one.1.x . 71 (3.4.1.5.7) Next.6) which.1. can be regarded as the electromagnetic current for the i → f electron transition.x . where Ni is the normalization constant. 2 Aµ = j(2) . (3.1. say a “spinless” muon. We can now rewrite the amplitude in a very suggestive form Z Tf i = −i jµf i eAµ d4 x (3. we just have to identify the electromagnetic potential Aµ with its source.1.x .8) . by comparison with (1.e− e− pC pA µ− µ− pD pB t-channel Figure 3. turned around by integration by parts. which determine the electromagnetic field Aµ associated with the current µ ej(2) = eNB ND (pD + pB )µ ei(pD −pB ). Using an analogous expression. This µ is done with the help of Maxwell’s equations. If the incoming e− has four momentum pi . for φf it follows that ejµf i = eNi Nf (pi + pf )µ ei(pf −pi ). using the results for the scattering of the “spinless” electron off an electromagnetic potential. we have φi (x) = Ni e−ipi . so that it acts on φ∗f Z Z ∗ µ 4 φf ∂µ (A φi ) d x = − ∂µ (φ∗f ) Aµ φi d4 x (3. the charged “spinless” muon. t → ±∞.4) where the surface term has been omitted as the potential is taken to vanish as |~x|.5) where ejµf i (x) = ie[φ∗f (∂µ φi ) − (∂µ φ∗f )φi ] (3. The Feynman diagram of such a process is shown in Fig. we can calculate the scattering of the same electron off another charged particle.5: Tree level diagram for electron-muon scattering.30). 3.

7).1. we find the tree level amplitude for electron muon scattering   Z −1 µ 4 (1) 2 j(2) d x . we must set the normalization N of the free particle wave functions (1.11)   gµν −iM = [−ie(pA + pC ) ] −i 2 [−ie(pB + pD )ν ] . (3. and carrying out the x integration we find. Recall that the probability density of particles described by φ is ρ = 2E|N|2 .4. M.10) jµ (x) Tf i = −ie q2 Substituting (3.1.1. Now.9) q with q = pD − pB . Whenever the same vertex or internal line occurs in a Feynman diagram the corresponding factor will contribute multiplicatively to the amplitude −iM for that diagram.28).x we obtain 1 µ Aµ = − 2 j(2) . Note that the multiplicative of the three factors gives −iM. Consequently. The four-momentum q of the photon is determined by four-momentum conservation at the vertices.1. The proportionality of ρ to E was just what we needed to compensate for the 72 .” Each vertex factor contains the electromagnetic coupling e and a four-vector index to connect with the photon index.12) shows that we would have obtained the same amplitude considering the muon scattering off the electromagnetic field Aµ produced by the electron. It is noteworthy that the photon propagator carries Lorentz indices because it is a spin-1 particle.1. Tf i = −iNA NB NC ND (2π)4 δ (4) (pD + pC − pB − pA ) M (3.where the momenta are defined in Fig.x = −q 2 eiq.1.8) and (3.11). Inserting this field due to the muon into (3. using 2 eiq.5). To relate these calculations to experimental observables. We see that q 2 6= 0.12) where µ A consistency check on (3.1. (3.5. and we say the photon is “virtual” or “off-mass shell.1. The delta function expresses energy-momentum conservation for the process. The particular distribution of the minus signs and factors i has been made up to give the correct result for higher order diagrams. q (3. is called the invariant amplitude.1. 3. as defined by (3.

and (2π)4 times the other gives T V . (The derivation of the formula for particle decay rates proceeds along similar lines. but we have 2E particles in V . of final states/particle = V d3 p .1. quantum physics restricts the number of final states in a volume V with momenta in element d3 p to be V d3 p/(2π)3. yielding No.17) Therefore. and 73 . for particles C. Therefore. V ρ dV = 2E. (initial flux) (3. Upon squaring.1.1. making use of (3.18) It is easiest to calculate the initial flux in the lab frame. This leads to the covariant normalization 1 (3.11).1. The number of beam particles passing through unit area per unit time is |~vA |2EA /V . one delta function remains. We then R work with a volume V and normalize to 2E particles within that volume. (2π)3 2E (3.14) where T is the time interval of the interaction and the transition amplitude is given in (3. No.1.1.16) where the factors in brackets allow for the “density” of incoming and outgoing states.) For a single particle. V4 (3.15) Experimental results on AB → CD scattering are quoted in the form of a “cross section. of available final states = V d 3 pC V d 3 pD .13) we obtain Wf i = (2π)4 δ (4) (pA + pB − pC − pD )|M|2 . D scattered into momentum elements d3 pC . V The transition rate per unit volume of the process A + B → C + D is |Tf i |2 Wf i = . d3 pD .” which is related to the transition rate according to cross section = Wf i (number of final states) .1. (2π)3 2EC (2π)3 2ED (3.13) N=√ . TV (3.1. see Appendix A.Lorentz contraction of the volume element d3 x and to keep the number of particles ρd3 x unchanged.

22) 4π 2 4EC ED √ where dΩ is the element of solid angle about p~C and s ≡ W = EA + EB . Now. (3.1.13) of the wave function is N = 1. (3. (3.1.23) = pf + dpf EC ED dQ = and rewrite Eq.1. For reactions symmetric about the collision axis. and (3. we obtain   1 1 dW .21) (2π)3 2EC (2π)3 2ED by partially evaluating the phase-space integrals in the center-of-mass frame. We first choose to integrate all three components of pD over the delta functions enforcing 3-momentum conservation.20) |M| δ (p +p −p −p ) A B C D |~vA |2EA 2EB V 4 (2π)6 2EC 2ED Note that the arbitrary normalization volume cancels. Consequently.22) as   1 1 pf dW dΩ δ(W − EC − ED ) dQ = 4π 2 4 EC + ED 1 pf √ dΩ .1. we obtain the normalization-independent measure of the ingoing “density” by taking 2EA 2EB .1. we can simplify the Lorentz invariant phase space factor dσ = d 3 pD d 3 pC .e.1.19) into (3.24) 4π 2 4 s 74 .1. (3. (3. = (3.15).19) Initial flux = |~vA | V V Substituting (3.18). we normalize to 2E particles/unit volume. and the normalization factor (3. This sets ~pC = −~pD and converts the Lorentz invariant phase space factor to the form dQ = (2π)4 δ (4) (pA + pB − pC − pD ) 1 d 3 pC 1 δ(EA + EB − EC − ED ) 4π 2 2EC 2ED 1 p2C dpC dΩ = δ(W − EC − ED ) .1. i. hereafter we drop V and work in unit volume. using W = EC + ED = (p2f + m2C )1/2 + (p2f + m2D )1/2 .1.16) we obtain the differential cross section dσ for scattering into d3 pC d3 pD 4 V4 d 3 pC d 3 pD 1 2 (2π) (4) . (3.1. Therefore.1.the number of target particles per unit volume is 2EB /V .. (3.1.

the incident flux for a general collinear collision between A and B reads.pB )2 − m2A m2B )]1/2 . F = |~vA − ~vB | 2EA 2EB = 4(|~pA |EB + |~pB |EA ) = 4[(pA . where |~pC | = |~pD | = pf . On the other hand. and hence the differential cross section in the center-of-mass is .e− e− pA pB e− e− e− pC pA pD pB e− pD e− t-channel e− pC u-channel Figure 3.6: Lowest-order Feynman diagrams for Møller scattering.

dσ .

.

1 pf |M|2 = .

1. 64π 2 s pi (3.26) where |~pA | = |~pB | = pi . (3.26) reduces to . In the special case where all four particles have identical masses (including the commonly seen limit m → 0). Eq.m.1.25) (3.1. dΩ c.

dσ .

.

27) = dΩ .1. |M|2 . (3.

The tree level invariant amplitude for the scattering of a spinless electron is then   2 e (pA + pC )µ (pB + pD )µ e2 (pA + pD )µ (pB + pC )µ − .c. 64π 2 s In closing we note that for electron-electron scattering we need to take into account that we have identical particles in the initial and final states.m. −iM = −i − (pD − pB )2 (pC − pB )2 (3. 75 . 3.1. we have two Feynman diagrams shown in Fig. Therefore.6.28) Note that the symmetry under pC ↔ pD ensures that M is also symmetric under pA ↔ pB . and hence the amplitude should be symmetric under the interchange of particle labels C ↔ D and A ↔ B.

where we have again taken e to be the charge of the electron. which satisfies the Dirac equation (γµ pµ − m)ψ = 0.6.32) 76 .x (3. the amplitude for the scattering of an electron from state ψi to state ψf is Z Tf i = −i ψf„ (x) V (x) ψi (x) d4 x Z = −ie ψ f γµ Aµ ψi d4 x Z = ie jµf i Aµ d4 x (3.2. The introduction of γ 0 is to make (3.31) can be regarded as the electromagnetic transition current between states i and f.2. The equation for an electron in an electromagnetic field Aµ is obtained by the substitution pµ → pµ −eAµ .x. Repeating the steps of the preceding section.2. )ψ = V ψ.132). so that the potential energy enters in the same way as in the Schr¨odinger equation.29) of the form (E + . .3. ψ = u(p)e−ip.2 Unpolarized Cross Section We have seen that a free electron of four-momentum pµ is described by a spinor.2. Using first-order perturbation theory (1. = −i(euC γµ uA ) q2 (3. the tree level transition amplitude for electron-muon scattering is   Z −1 µ 4 (1) 2 jµ (x) Tf i = −ie j(2) d x q2   −1 (euD γ µ uB ) (2π)4 δ (4) (pA + pB − pC − pD ) . (3.29) where the perturbation is given by γ 0 V = eγµ Aµ .2. We find (γµ pµ − m)ψ = γ 0 V ψ. .30) where e jµf i ≡ e ψ f γµ ψi = e uf γµ ui ei(pf −pi ).

To allow for scattering in all possible spin configurations. we must amend the cross section formulae of Sec. with pA = k.2. To obtain the (unpolarized) cross section. pC = k ′ .34) To calculate the unpolarized cross section. 3. we therefore have to make the replacement |M|2 → |M|2 ≡ X 1 |M|2 .2.38) . That is. 77 (3.2. pB = p. we average over the spins of the incoming particles and sum over the spins of the particles in the final state.33) and so we have µ −iM = (−ieuC γ uA )  −igµν q2  (−ieuD γ ν uB ) .35) as |M|2 = e4 µν (µ) L L q 4 (e) µν where the tensor associated with the electron vertex is 1 X [u(k ′ )γ µ u(k)][u(k ′ )γ ν u(k)]∗ . Lµν ≡ (e) 2 e−spins (µ) and with a similar expression for Lµν . Recall that the invariant amplitude M is defined by Tf i = −i(2π)4 δ (4) (pA + pB − pC − pD ) M .36) M = −e u(k ) γ u(k) 2 u(p′ ) γ ν u(p) q and then carry out the spin sums (the momenta are defined in Fig.2.1. and q = k − k ′ ). sB are the spins of the incoming particles. (2sA + 1)(2sB + 1) spins (3.35) where sA . It is convenient to separate the sums over the electron and muon spins by writing (3. (3. 3. (3.37) (3.2.where q = pA − pC . pD = p′ .2.2. we have to take the square of the modulus of   1 ′ µ 2 (3. and the factor (2π)4 times the delta function arises from the integration over the x dependence of the currents. By unpolarized we mean that no information about the electron spins is recorded in the experiment.5.

78 . (3. p )(k .2. p′ ) q4 + m2e p′ .41) (e) = Tr[(6 k + me ) γ (6 k + me ) γ ] . . . . p) + (k ′ . with summation over repeated indices implied) 1 X (s′ ) ′ µ X (s) (s′ ) ′ ν uβ (k) u(s) uα (k ) γαβ (3. we note that the second square bracket of (3. the complex conjugation in (3.2.38) (a 1 × 1 matrix for which the complex and hermitian conjugates are the same) is equal to [u„ (k ′ ) γ 0 γ ν u(k)]„ = [u„ (k) γ ν „ γ 0 u(k ′)] = [u(k)γ ν u(k ′ )] . k + 2m2e m2µ ] .The spin summations look like a forbidding task.2.43) where mµ is the mass of the muon. To begin.2. .k − m2e )g µν ) . 2 A straightforward evaluation of the tensor associated with the electron vertex (3. wellestablished trace techniques considerably simplify such calculations.2. Fortunately. That is. Forming the product of (3. p − m2µ k ′ . (3. yielding (µ) Lµν = 2(p′µ pν + p′ν pµ − (p′ . we finally arrived at the following “exact” form for the spin average e− µ− → e− µ− amplitude. (3.38) simply reverses the order of the matrix product. |M|2 = (3.40) Lµν = γ (k) γγδ uδ (k ) .39) where we have used γ ν „ γ 0 = γ 0 γ ν .38) explicitely in terms of individual matrix elements (labeled α. Lµν (e) becomes the trace of the product of 4 × 4 matrices 1 ′ µ ν Lµν (3.2.2. 8e4 ′ ′ [(k .2. (e) 2 ′ s s | {z } (6k +me )βγ where me is the mass of the electron. p)(k .44) In the extreme relativistic limit. We now write the complete product in (3.43).42) The evaluation of Lµν (µ) is identical.p − m2µ )gµν ) .2.41) using the trace theorems given in Appendix B leads to 1 1 Tr(6 k ′ γ µ 6 kγ ν ) + m2e Tr(γ µ γ ν ) Lµν (e) = 2 2 ′µ ν ′ν µ = 2(k k + k k − (k ′ .2. we could neglect the terms containing m2e and m2µ . Thus.42) and (3. β.2.2.

However. −~kf ). pB . D) and pA + pB = pC + pD due to energy momentum conservation. ~ki . pC . Mandelstam. Rather than these. s = (pA + pB )2 = 4(k 2 + m2 ). and thus invariant variables are the scalar products pA . pA . 112.. pB = (E. t.45) u = (pA − pD )2 = (pB − pC )2 . −~ki ). ~kf = k 2 cos θ. pA = (E. For any “two particle to two particle” process (A+B → C +D) we have at our disposal the four-momenta associated with each particle. 1344 (1958). In the center-of-mass system for the reaction A+B → C +D. Phys.(pB − pC − pD ) i = X m2i . and θ is the center-of-mass scattering angle.. it is conventional to use the related (Mandelstam) variables1 s = (pA + pB )2 = (pC + pD )2 t = (pA − pC )2 = (pB − pD )2 (3. pA . s is equal to 2 .46) i i. 79 . and u let us evaluate them explicitly in the center-of-mass frame for particles all of mass m. because p2i = m2i (with i = A. pC = (E. To get a better feel for s.3. only two of the three variables are independent. we have t ≤ 0 and u ≤ 0. we have s ≥ 4m2 . C. Note that t = 0 (u = 0) corresponds to forward (backward) scattering.3. B.e. As k 2 ≥ 0. pD = (E. Rev.e. E = (k 2 + m2 )1/2 .3. where Ecm is the sum of the energies of the square center-of-mass energy Ecm 1 S. ~kf ). (3. t = (pA − pC )2 = −(~ki − ~kf )2 = −2k 2 (1 − cos θ) u = (pA − pD )2 = −(~ki + ~kf )2 = −2k 2 (1 + cos θ) where.3 Mandelstam Variables In high energy physics. ~ki). pD . X s+t+u = m2i + 2p2A + 2pA . and since −1 ≤ cos θ ≤ 1. cross sections and decay rates are written using kinematic variables that are relativistic invariants. i.

This is called the s-channel process. whereas s ≤ 0 and u ≤ 0 are squares of momentum transfers.49) . pC ≡ k ′ . p′ ≃ −2k ′ . The required interchange is k ′ ↔ −p. s ↔ t in (3. and u = (pA − pD )2 .47) u ≡ (k − p′ )2 ≃ −2k .3. in the s-channel s is positive. p ≃ 2k ′ .3. u is positive and represents the square of center-of-mass energy of the AD system. and we obtain |M|2 |M|2 = 2e4 80 t2 + u2 .48). (3. here s = (pA − pB )2 . This is called the t-channel process. and pD ≡ p′ . and u = (pA + pD )2 .44) can be rewritten as 8e4 [(k ′ . p′ )] = ′ 4 (k − k ) s2 + u 2 = 2e4 . s = (pA − pB )2 . the unpolarized e− µ− → e− µ− scattering amplitude (3. From this process we can form another process. The antiparticles have four-momenta which are the negatives of the momenta of the particles: pB → −pB and pC → −pC relative to the s-channel reaction.3. k ′ ≃ −2p . A + D → B + C. At high energies. In the extreme relativistic limit the Mandelstam variables become s ≡ (k + p)2 ≃ 2k . and u (which is not an independent variable) represents the square of the momentum transfer between particles A and D. p′ ≃ −2k 2 (1 − cos θ) . pB ≡ p. s2 (3. where pA ≡ k. that is. In this channel t is positive and represents the square of center-of-mass energy of the AC system. p′ )(k . (3. t = (pA + pC )2 . p′ ≃ 4k 2 .48) t2 We may also obtain the amplitude for e− e+ → µ+ µ− by “crossing” the result for e− µ− → e− µ− . AC → B + D. Correspondingly here. This is called the u-channel process. t represents the square of the momentum transfer between particles A and C. p)(k . p ≃ −2k 2 (1 + cos θ) . by taking the antiparticle of C to the left-hand side and the antiparticle of B to the right-hand side. p) + (k ′ . by taking the antiparticle of B to the left-hand side and the antiparticle of D to the right-hand side. We can form yet another process from the above. As we have seen.particles A and B.2. t = (pA − pC )2 . In this channel. while t and u are negatives. while s ≤ 0 and t ≤ 0 are squares of momentum transfers.3. Hence. t ≡ (k − k ′ )2 ≃ −2k .

3. where now e− e+ → µ+ µ− is the s-channel process.7: Feynman diagram for e+ e− → µ+ µ− .7.1.e− µ− k k′ p p′ µ+ e+ s-channel Figure 3. This result can be translated into a differential cross section for e− e+ → µ+ µ− scattering using (3. The corresponding tree level diagram is drawn in Fig.27). In the center-of-mass frame we have .

1 dσ .

.

3. (3.50) . = 2e4 [ 21 (1 + cos2 θ)] .

this becomes . Using α = e2 /4π. dΩ cm 64π 2 s where the quantity in square brackets is (t2 + u2 )/s2 .

dσ .

.

3. (3. α2 = (1 + cos2 θ) .51) dΩ .

(3. Phys.53) H.52) A comparison of these results with PETRA data2 is shown in Figs.3. where Ebeam is the energy of each beam. Phys.cm 4s To obtain the reaction cross section. resulting in e+ e− inter√ actions with s = 2Ebeam . 3s (3. 163 (1989).9. Phys. B 161. Equation (3. Behrend et al. 209 (1987). J. 283 (1982). 3. we integrate over θ and φ σe+ e− →µ+ µ− = 4πα2 . [JADE Collaboration]. Bartel et al. The PETRA accelerator consists of a ring of magnets which simultaneously accelerate an electron and a positron beam circulating in opposite directions.3. Z. W. Lett. Lett. [CELLO Collaboration]. B 191. 507 (1985).8 and 3. Lett. In selected spots these beams are crossed. Phys. 81 . C 14. C 26. Z.3.52) can be written in numerical form as σe+ e− →µ+ µ− = 20(nb) 2 /GeV2 Ebeam 2 . Phys. 188 (1985). B 222.

.and u-channel diagrams drawn in Fig. corrections to (3. .3. α4 .√ Figure 3. and so the amplitude should be symmetric under the interchange of particle labels C ↔ D (and A ↔ B). The dot-dashed line shows the relativistic limit of lowest order QED prediction. for e− e− → e− e− . we can simply use the antiparticle prescription to “cross” the result for e− e− → e− e− .e. 3. we have identical particles in the initial and final states. one can immediately check by 82 . the amplitudes of higher order diagrams. To obtain the amplitude for e− e+ → e− e+ . arising due to interference with. . i. (3. Furthermore..2) to calculate the (lowest-order) amplitude for Møller scattering. There are. or directly from. we have to calculate the t.6. of course. We can now use the procedure sketched in Sec. .53) of order α3 . As noted in the analysis of spinless electrons.8: The e+ e− → µ+ µ− angular distribution for h si = 39 GeV.

Figure 3. The dot-dashed line shows the relativistic limit of lowest order QED prediction.9: Solid (open) symbols indicate the cross section for e+ e− → µ+ µ− (e+ e− → τ + τ − ) measured at PETRA versus the center-of-mass energy. 83 .

10 and 3.and u.1 we give the amplitudes for all these processes in the extreme relativistic limit. the momentum carried by the virtual particle. i. 84 . corresponding to t.γ γ k k′ p p′ e− e− e− e− p′ p γ γ k′ k u-channel s-channel Figure 3. inspection of Figs.and u-channel exchanged with photons and electrons being almost on mass shell.. Recall that t. γ e− p p k′ k p γ e+ k γ e− ′ p′ γ e+ t-channel k′ u-channel Figure 3. When the mediator has a very small momentum squared (i.e. and make the replacement α → αs . In fact.11: Lowest-order Feynman diagrams for pair annihilation. in addition to their spins. The origin of the forward and backward peaks in the differential cross section is identified.e. Interaction with small deflections thus occurs with large cross sections.are the squares of the three momentum transferred. 3. the results are identical except that we must average (sum) over the color of the initial (final) quarks.10: Lowest-order Feynman diagrams for Compton scattering. then by the uncertainty principle the range of interaction is very large. In Table 3. q q¯ → q q¯ interactions via single gluon exchange.. where αs = gs2 /4π is the quark gluon coupling. almost on its mass shell).11 that a similar analysis applies to obtain the amplitude of pair annihilation e+ e− → γγ by crossing the amplitude for Compton scattering e− γ → e− γ. Similar results are found in QCD for the strong qq → qq.

1: Leading order contributions of some QED processes.Table 3. |M|2/2e4 Process s2 + u 2 t2 } | {z Møller scattering | {z } e− e− →e− e− + s2 + t2 2 | u {z } + interference forward (Crossing s ↔ u) 2s2 tu |{z} backward (u ↔ t symmetric) s2 + u 2 t2 } | {z Bhabha scattering {z } | e− e+ →e− e+ forward s2 + u 2 t2 } | {z e− µ− → e− µ− + 2u2 ts |{z} interference u2 + t2 + s2 } | {z timelike forward u2 + t2 s2 } | {z e− e+ → µ− µ+ timelike Compton scattering | {z } − e− γ→e− γ s u − s u |{z} |{z} timelike backward t u + t u |{z} |{z} pair annihilation | {z } e+ e− →γγ forward 85 backward .

. For example.56) is i(22 + m2 )φ = −iV φ . 440 (1950). as the perturbation parameter. H0 |ni = En |ni. 20. .4. Phys. . Feynman.55). We saw earlier [Eq. and the propagator may thus be regarded as i times the inverse of the Schr¨odinger operator.e. Tf i = 2πδ(Ef − Ei )hf |(−iV ) + (−iV ) −i(Ei − H0 )ψ = −iV ψ (3. .142)] that the nonrelativistic perturbation expansion of the transition amplitude is # " X 1 hf |V |ni hn|V |ii + . rather than V . which leads to a time dependence e−iV t in the interaction picture.6.55) Ei − H0 P where we have made use of the completness relation |nihn| = 1. It is natural to take (−iV ). P.54) where we have associated factors of hf |V |ni with the vertices and identified 1/(Ei − En ) as the propagator. Formally we may therefore rewrite (3. 80. 367 (1948). Rev.. (3. Tf i = −i2πδ(Ef − Ei ) hf |V |ii + Ei − En n6=1 (3. the form of the Klein-Gordon equation corresponding to (3. The state vectors are eigenstates of the Hamiltonian in the absence of V . The −i arises from the i in i∂ψ/∂t = V ψ.4.1).54) as i (−iV ) + .56) acting on the intermediate state. We can now apply the same technique to various relativistic wave equations to deduce the form of the propagators for the corresponding particles.4.3 Our primary aim is to motivate Feynman rules and to calculate physical amplitudes. Mod. (1. (3.4 Feynman Rules This section encompasses a heuristic treatment of QED based on Feynman’s intuitive space-time approach. 4 86 .3.4. Phys.4. Guided by the relativistic generalization of (3. |ii . i. . Rev.1.57) see (3.4.4.4 That is. the vertex factor is −iV . we expect the propagator for a spinless particle to be the inverese of the operator on 3 R.

4.4.60) where we have used p/p/ = p2 and the completeness relation (1.89).4. The propagator for the photon is not unique.4.59): X 1 i i(p/ + me ) i = uu .59) As before.4. we must multiply by −i. We would also integrate over the different momentum states that propagate.4. The numerator contains the sum over the spin states of the virtual electron. on account of the freedom in the choice of Aµ . The electron propagator is therefore the inverse of −i times the left-hand side of (3. this momentum is fixed by the momenta of the external particles. For an intermediate state of momentum p. the vertex factor is is −ieγ µ . an electron in an electromagnetic field satisfies (p/ − me )ψ = eγ µ Aµ ψ .the left-hand side of (3. Hence.61) p2 − m2 spins The spin sum is the completeness relation. this gives i 1 = 2 .4) can be written as (g νλ 22 − ∂ ν ∂ λ )Aλ = j ν 87 (3.63) The wave equation for a photon (1. For the diagrams we have considered so far.58) 2 2 i(−p + m ) p − m2 In a similar fashion.4. (3. (3. we include all possible spin states of the propagating particle.4. the general form of the propagator of a virtual particle of mass m is X i . In summary.2. (3.64) .57). (3.5. = 2 = p − m2e p2 − m2e s p/ − me −i(p/ − me ) (3. (3.4. Recall that physics is unchanged by the transformation that is associated with the invariance of QED under phase or gauge transformations of the wavefunctions of charged particles Aµ → A′µ = Aµ + ∂µ χ .62) where χ is any function that satisfies 22 χ = 0 .

4. In such a case.4. From (3.4.69) for A and B. which is the quantity in brackets on the right-hand side of (3. (3.69) multiplied by i. is found to be i(−g µν + pµ pν /M 2 ) .66) where the four vector ǫµ is the polarization vector of the photon.64) we see that the wavefunction Bλ for a free particle satisfies  νλ 2  g (2 + M 2 ) − ∂ ν ∂ λ Bλ = 0 .68) Proceeding exactly as before. The propagator.4.4. In our discussion so far. the wave equation (3. a photon propagator cannot exist until we remove some of the gauge freedom of Aλ . 88 (3. With this in mind. x .4. we have chosen to work in the Lorentz class of gauges with ∂λ Aλ = 0.4.68).70) We can show that the numerator is the sum over the three spin states of the massive particle when taken on-shell p2 = M 2 . the wavefunction Aµ for a free photon satisfies the equation 22 Aµ = 0 . Two terms cancel and we find M 2 ∂ λ Bλ = 0 .4.and. in fact.65) which has solutions Aµ = ǫµ (q)e−iq .67) The wave equation for a spin-1 particle of mass M can be obtained from that for the photon by the replacement 22 → 22 + M 2 . ∂ν .4.64) simplifies to g νλ 22 Aλ = j ν . (3. we determine the inverse of the momentum space operator by solving  νλ −1 g (−p2 + M 2 ) − pν pλ ) = δλµ (Agµν + Bpµ pν ) (3.4. of (3.71) . p2 − M 2 (3. q2 (3.4. and since gµν g νλ = δµλ (where δµλ is the Kronecker delta). We first take the divergence. the propagator (the inverse of the momentum space operator multiply by −i) is i −gµν . (3.

4.4.4.4.73) with free particle solutions . Nevertheless. The completeness relation (in an obvious notation) 89 .4. for a photon traveling along the z-axis. Likewise. In this case we can explore the consequences of the additional gauge freedom (3. ∂µ Aµ = 0 gives. 1. we may take ǫ1 = (1. the wave equation reduces to (22 + M 2 )Bµ = 0 (3.Hence for a massive vector particle.x (3. 0). we can anticipate that it is associated with a particle of spin-1.4. 0.62).71) demands pµ . the Lorentz condition for photons. This means that there are only two independent polarization vectors and they are both transverse to the three-momentum of the photon. For example. ǫ2 = (0.4. reducing the number of independent components of ǫµ to three. where −gµν implies we are summing over four polarization states.63) is satisfied.72) Bµ = ǫµ e−ip (3. ǫ0 ≡ 0 and the Lorentz condition reduces to ~ǫ . This (noncovariant) choice of gauge is known as the Coulomb gauge. ǫµ = 0 and so reduces the number of independent polarization vectors from four to three in a covariant fashion.66) into (3. The condition (3. it is not a gauge condition. we have no choice but to take ∂ λ Bλ = 0. x . Substituting this. Choose a gauge parameter χ = iae−iq. (3. (3.75) In other words. ǫ′µ ) which differ by a multiple of qµ describe the same photon. two polarization vectors (ǫµ .4. As a consequence. qµ .76) A free photon is thus described by its momentum q and a polarization vector ~ǫ.4. we have associated with a virtual photon the covariant propagator i(−gµν )/q 2 . 0) .4. Since ~ǫ transforms as a vector. together with (3. ǫµ = 0.74) with a constant so that (3. ~q = 0.62) shows that the physics is unchanged by the replacement ǫµ → ǫ′µ = ǫµ + aqµ . We may use this freedom to ensure that the time component of ǫµ vanishes.

For a real photon. 90 . How can one reconcile the two descriptions? Consider a typical Feynman diagram containing a virtual photon exchanged between charged particles.4. we can obtain the invariant amplitude M by drawing all (topologically distinct and connected) Feynman diagrams for the process and assigning multiplicative factors (summarized in Table 3. (3. q 0 ≈ |~q|. leaving only the two transverse contributions. then j3 = j0 and the longitudinal and scalar contributions cancel each other. 0. For both the A and B currents this implies q µ jµ = q 0 j0 − |~q|j3 = 0 . Fig. (3.77) scalar However. For such diagrams (e. we can therefore make the replacement X = ǫTµ ∗ ǫTν → −gµν . | {z } (3.2) with the various elements of each diagram.4. being emitted and then sooner or later being absorbed. 0. for a virtual photon the longitudinal and scalar components cannot be neglected. in the spirit of (3.78) longitudinal/scalar where we have taken the photon four-momentum q µ = (q 0 .4. |~q|).80) T On the other hand.55). we choose the 3-axis to be along qˆ..5) we have found a transition amplitude of the form  µν  Z −g A 2 jνB (x) d4 x jµ (x) Tf i = −ie q2   Z  A B   j1 j1 + j2A j2B j3A j3B − j0A j0B  4 = −ie +   d x.is given by −gµν = = 4 X ǫ(λ)∗ ǫν(λ) = µ X ǫTµ ∗ ǫTν + T λ=1 (δij − qˆi qˆj ) | {z } transverse X L ǫL∗ µ ǫν + L + qˆi qˆj |{z} longitudinal + X S ǫS∗ µ ǫν S (−gµ0 gν0 ) .79) Therefore if the exchange photon is almost real. 3. Recall that charge conservation gives rise to the continuity equation ∂ µ jµ = 0.4. q2 q2   | {z } | {z } 2 transverse (3. in a sense every photon is virtual.4. That is. Now.g.

out) photon (in.2: Feynman rules for −iM. • Identical fermions: −1 between diagrams which differ only in e− ↔ e− or initial e− ↔ final e+ .12 fermion i(6p+m) p2 −m2 massive spin-1 boson −i(gµν −pµ pν /M 2 ) p2 −M 2 massless spin-1 boson (Feynman gauge) −igµν p2 • Vertex Factors photon−spin-0 (charge e) −ie(p + p′ )µ photon−spin. v ǫµ .12 spin.Table 3. u v.12 spin-1 boson (or antiboson) fermion (in. Multiplicative Factor • External Lines spin-0 spin. 91 . include −1 if fermion loop and take the trace of associated γ-matrices. out) antifermion (in. out) 1 u. ǫ∗µ • Internal Lines − Propagators spin-0 boson i p2 −m2 spin.21 (charge e) −ieγ µ R • Loops: d4 k/(2π)4 over loop momentum.

K. Webber. we attempt to provide a glimpse of the rich structure of QFT and expose the reader to the concepts of loops. Cosmol. An Introduction to Quantum Field Theory.5. R. 1995). . σe+ e− →µ− µ+ q (3. and hence X σe+ e− →qq¯ σe+ e− →hadrons = q X e2q σe+ e− →µ− µ+ . Ellis.5. Nonetheless. To obtain the cross section for producing all types of hadrons. Schroeder.. which can be found in most field theory books. Phys.82) where we have taken account of the fractional charge of the quark. W. the following discussion is rather incomplete and a few results are not explicitly derived. only unrevealing algebra is omitted. . eq . (3.84) = 3 q 5 E. QCD and collider physics. (3. 4πα2 . renormalization. Because QFT is not the main subject of this course. 92 . Monogr. and running couplings in a concise and physical way. Nucl.5. The cross section for the (QED) process e− e+ → q q¯ is readily obtained from that for the process drawn in Fig. d. we must sum over all quark flavors q = u.83) This simple calculation leads to the dramatic prediction X σe+ e− →hadrons R≡ =3 e2q . the center-of-mass energy squared is s = Q2 = 4Ebeam . The extra factor of 3 arises because we have a diagram for each quark color and the cross sections have to be added. J.81) σe+ e− →µ+ µ− = 3Q2 a result obtained in (3.g. R.52). (Addison-Wesley. E. Part. 8. . Peskin and D. Reading. M. Phys. 3. Stirling and B. Camb. s.5 Beyond the Trees In this section. 1 (1996).3.7.3.5 The bulk of hadrons produced in e− e+ annihilations are fragments of a quark and antiquark produced by the process e− e+ → q q¯. . V.5. The required cross section is σe+ e− →qq¯ = 3 e2q σe+ e− →µ− µ+ (3. Here.

Bernardini et al. Phys.84) as a function of the total e+ e− center-of-mass energy. 200 (1974). d. T.. c. [CLEO Collaboration].6 The 6 See e. 969 (1982).5.  2 for u.g. 681 (1986). 1339 (1990). c. D. D 34.) Because σe+ e− →µ− µ+ is well known (see Fig. Z. s. D 42. Kumita et al. Lett.12: Ratio R of (3. (3. We have h  2 2 i 2 R = 3 32 + 31 + 13 = 2 for u. s. s. d. Rev. J. B. 93 . [AMY Collaboration]. Phys. 3. Adeva et al. Lett. 3. their flavors. [Mark-J Collaboration].85) = 10 3 3 In Fig. 381 (1985). Phys. 54. d. Rev.9). [TASSO Collaboration].. b . M. Besson et al. M. = 2 + 3 32 = 10 3  2 + 3 31 = 11 for u.12 these predictions are compared to the measurements of R.. (The sharp peaks correspond to the production of narrow resonances just below or near the flavor thresholds. D 26. 307 (1984).5. Phys. Rev. as well as their colors. a measurement of the total e+ e− annihilation cross section into hadrons therefore directly counts the number of quarks.  R 0 10 3 10 2 J= ! 10 0 00 Z   0  1 -1 10 1 10 10 2 Q (GeV) Figure 3. Phys. Althoff et al. B 51. Siegrist et al. C 22. Phys. Rev.

5.83) is based on the (leading order) process e+ e− → q q¯. 94 .value of R ≃ 2 is apparent below the threshold for producing charmed particles at Q = 2(mc + mu ) ≃ 3.86) R(α.  e As always. + .5. Equation (3. (3. we must expect that for the dimensionless observable R. because R = 11 3 by a factor of 3 if there was only one color. we should also include diagrams where the quark and/or antiquark radiate gluons.7 GeV. However. in (3.88) the antiparticles are drawn using only particle (e− . Hereafter we will adopt this simplified notation whenever there is no danger of confusion. These results for R will be modified when interpreted in the context of QCD. R = 11 as predicted.5.5. s) = σe+ e− →µ+ µ− is a function of the electromagnetic coupling α. R(α. These measurements con3 would be reduced firm that there are three colors of quark. This prediction disagrees with experiment and is. 4π e (3. not true in renormalizable QFT. When the annihilation energy far exceeds the light masses m of quarks and leptons. In general σe+ e− →qq¯ (3.89) because there is no intrinsic scale in theories with massless exchange bosons.88) q.87) 2 and the annihilation energy s = 4Ebeam : e+ q. q) lines.5. s) −→2 constant s≫m (3.5. Above the threshold for all five quark flavors (Q > 2mb ≃ 10 GeV). e α= e e2 . but note that we omit the arrow lines indicating the time direction of the antiparticle’s four-momenta. in fact. The exchange of a massless photon is ultraviolet divergent. µ− .

Λ This seems ugly.5. (3.97) In other words. dΛ2 ∂z z=1 7 (3. 95 .5. Because α(s) is dimensionless.   R α(s). physical observables (e.g. we obtain    s  ∂ ∂ R α(s). Λ2 ) is of the form  s  (3.   ∂F 2 dα(s) Λ = (α(s). Thus.5. 2 .5. or via the coupling α. (3. 2 = 0 .94).. which can be written more explicitly: ∂R ∂α ∂R Λ2 2 + Λ2 2 = 0.91) This is the renormalization group equation. (3. (3. because any Λ-value is arbitrary.5. a scale is introduced into the calculation and the dimensionless observable R(α. Λ2 dR = 0. s.requiring the introduction of a cutoff Λ.93) − +β ∂t ∂α Λ where ∂α ∂α = .5. R) cannot depend on Λ.96) Λ which is consistent with (3.95) β = Λ2 in which the observable depends on s only via the coupling. dimensional analysis requires  s  α(s) = F α(Λ2 ).5. Equation (3. Using Λ2 ∂/(∂Λ2 ) = −∂/[∂ ln(s/Λ2 )]. (3. the renormalization group equation has the very simple solution.7 Therefore. it is not: Λ appears order by order in the perturbative series but not in the final answer.92) ∂Λ ∂Λ ∂α which exhibits that R can depend on Λ directly. dΛ2 (3.92) can be rewritten in the variable t ≡ ln(s/Λ2 ).94) 2 ∂Λ ∂t With the identification Λ2 = s. 1 . 2 .5.5.90) R = R α. z) = β(α) .5.

5.99) where k e0 Π(q 2 ) = e0 q q . explicit calculation (see Appendix C) confirms this and we therefore write Π(q 2 ) in terms of a divergent and finite part e20 Π(q ) = 12π 2 2 Z Λ2 dk 2 e20 −q 2 − ln 2 12π 2 m2e m2e k   e20 Λ2 = . 1 + Π(q 2 ) (3. (3.5.5. + e0 e0 = e20 − e20 Π(q 2 ) + e20 Π2 (q 2 ) − · · · . which can be computed perturbatively.5. ln 12π 2 −q 2 96 (3.101) .98) The “running” of the coupling is described by the β-function.. which is made explicit in order to introduce the summation (3.99).. In field theory the interaction of two electrons by the exchange of a virtual photon is described by a perturbative series e e0 e0 e0 q 2 + e+ e0 e e +. Π(q 2 ) is ultraviolet divergent as k → ∞. e20 = . We discuss this next.100) k+q Note the negative sign associated with the fermion loop. β(x) (3.5.The solution is t = ln(s/Λ) = Z α(s) α(Λ) dx .

103) e0  = ee0 = e 1 + 1 2 e 97 +   at q2= 2 .5.5.104) has been obtained by substituting the renormalized charge e for the bare charge using (3. – M = e  + = e 1 2 2 e0  at +. e(µ = 0) is the Thomson charge with α = e2 (µ = 0)/(4π) = 1/137....5.103) We never said what e0 was. This operation is performed at some reference momentum µ.5.g. in fact.5.102) .035999679(94). or e = e0  1 1 − Π(−q 2 = µ2 ) + · · · 2  (3. infinitesimal and combines with the divergent loop Π to yield the finite. (3. To illustrate how this works we calculate e− e− scattering.104) where (3.5. – q 2 = 2 (3. The amplitude is (ignoring identical particle effects) e0 + . e.The trick is to introduce a new charge e which is finite:   e2 = e20 1 − Π(−q 2 = µ2 ) + · · · .105) .. It is. physical charge e. (3.

q 2 = 2  2   2 α Λ Λ α ln − ln 2 3π −q 3π µ2  2  µ α ln = finite! = 3π −q 2 (3. 1 Q2 1 − = −b ln . Therefore (3. The ultraviolet cutoff is eliminated by renormalizing the charge to some measured value at Q2 = µ2 . cannot be calculated. The running charge (3.5.In the last term of (3. In a renormalizable theory this cancellation happens at every order of perturbation theory.109) . In summary.5.108) 2 1 − bα0 ln Q Λ2 with b = 1/3π.5. (3.102) can be written as   α0 . The electron charge.5.5.104) can be rewritten as: e M =   – – { at +.107) α = α0 1 − Π(q 2 ) + · · · = 1 + Π(q 2 ) For the QED result (3...106) The divergent parts cancel and we obtain a finite result to O(α2 ). α(Q2 ) α(µ2 ) µ2 98 (3.5.101).102) the perturbation series using infinitesimal charges e0 and infinite loops Π has been reshuffled order by order to obtain finite observables. α0 α(Q2 = −q 2 ) = (3. The price we have paid is the introduction of a parameter α(µ2 ) which is fixed by experiment. unfortunately.5.5. by using the substitution (3.104) we can just replace e0 by e as the additional terms associated with the substitution (3.105) only appear in higher order.5.5.

coupling α ≡ e gs g 1 g0 2 g2 4π b-value 1 3π g 2nq − 33 12π W i 4ng + 12 nd − 22 12π n + 12 nd − 20 3 g 12π B nq : number of quarks (2–6) ng : number of generations (3) nd : number of Higgs doublets (1) The formal arguments have revealed the screening of the electric charge. From Eq.94) and (3. In quantum field theory a 99 .95) it is clear that much of the structure of the gauge theory is dictated by identifying the momentum dependence of the couplings.99).One also notices that b determines the β-function to leading order in perturbation theory.5.5. There is physics associated with Eq.110) β(α) = ∂t In Table 3.3: b-values for the running of the coupling constants.5. (3.108) that ∂α(Q2 ) = bα2 + O(α3 ) .5. g ′ and the strong color charge gs . (3. We obtain indeed from (3. Table 3. (3.3 we have listed the b-values determining the running of the other standard model couplings: the weak couplings g.5.

5.109). On the other hand.  2 1 1 mZ 1 . negative b-value of −33/12π. (3.8 We note that. Rev. presumably leading to the confinement of quarks and gluons. 054008 (1999). Therefore α−1 (µ2 = 0) ≃ 137 is smaller than the short-distance value α−1 (µ2 = m2Z ) = 127.016. Phys. Erler. 100 .5. D 59. qualitatively. loops with gluons reverse that effect with a larger. While q q¯ pairs screen color charge just like e+ e− pairs screen electric charge (the 2nf /12π term in b). 8 J. The color charge grows with distance yielding the asymptotic freedom property: αs → 0 as Q → ∞. For 3 generations of quarks the b-value for QCD is negative.925 ± 0. the theory becomes strongly coupled (infrared slavery) at the energy scale Q2 ∼ Λ2QCD .charge is surrounded by virtual e+ e− pairs which screen the charge more efficiently at large than at small distances.111) − ≃9≃ ln 2 α(0) α(mZ ) 3π m2e see (3.

102. that is. D. Q2 ≡ −q 2 . the Landau pole of QCD corresponds to a resolution of nucleon’s size (somewhat below 1 fm or 10−15 m) and is referred to as the onset of the “deep inelastic regime. corresponding to 2 Q2 = MPl ≈ 1038 GeV2 . which is if the momentum transfer decreases towards the region ∼ the natural habitat of nucleons and pions. Y. deep inelastic scattering experiments paved the way for understanding the structure of the nucleon. which defines the “Landau pole” of QCD. 489 (1955). Cooper-Sarkar. In particular.1 This behavior is totally different from QED. 2004). This theory is asymptotically free. Deep Inelastic Scattering. Dokl. Landau and I. 2 101 . Nauk Ser. Pomeranchuk.”2 In the late 60s. (Oxford University Press. it can be treated in a perturbative way for very large values of the four-momentum transfer.1 Deep Inelastic Scattering Hadrons are composite systems with many internal degrees of freedom. or 10−35 m. Akad. for which α(Q2 ) diverges for huge momentum transfers at the Planck scale. below any distance ever to be resolved by experiment. Devenish and A.Chapter 4 Hard Scattering Processes 4. is expected to diverge if Q2 decreases to values near Λ2QCD ≈ (250 MeV)2 . the so-called “partons” are described by QCD. Contrariwise. Fiz. the “running” of the QCD coupling constant αs (Q2 ). The strongly interacting constituents of these systems. However. the binding forces become increasingly strong < 1 GeV. R. When trying to deduce the struc1 L.

The procedure to obatin this information is to scatter electrons on this cloud. 4. the underlying idea is quite simple and straightforward.1: Lowest-order electron scattering by a charge cloud.1. Suppose we want to determine the charge distribution shown in Fig. ture of composite objects. like hadrons. this would give us a form factor F (q). be the cloud of an atom. for example. measure the angular cross section and compare it with the known cross section for scattering of a point distribution. As the charge cloud certainly is not a point charge. .e− kf ki e− Figure 4.e.. which could. i.

dσ dσ .

|F (q)|2 ,
(4.1.1)
=
dΩ
dΩ

point

where q is the momentum transfer between the incident electron and the
target, q = ki − kf . We then attempt to deduce the structure of the target
from the F (q) so determined.
We can gain insight into this technique by first looking at the scattering
of unpolarized electrons of energy E from a static spinless charge distribution
−Zeρ(~x), normalized so that
Z
ρ(~x) d3x = 1 .
(4.1.2)
For a static target, it is found that the form factor in (4.1.1) is just the
Fourier transform of the charge distribution
Z
F (~q) = ρ(~x) ei~q . x~ d3 x ,
(4.1.3)

while the reference cross section for a structureless target (see Appendix D)
is

dσ .

(Zα)2 E 2
[1 − v 2 sin2 (θ/2)] ,
(4.1.4)

=
dΩ

point dΩ .

Mott 4k 4 sin4 (θ/2) 102 .

ρ(~x) d3 x F (~q) = 1 + i~q . then we already know the answer. ~x − 2  Z  1 2 2 2 = 1 + iq r cos θ − q r cos θ .6) of the charge cloud. .1 is soft and with its large wavelength can resolve only the size of the charge distribution ρ(r) and is not sensitive to its detailed structure.where k = |~ki| = |~kf |. a function of r ≡ |~x| alone. Z hr 2 i = r 2 ρ(r) 4π r 2 dr . The small-angle scattering therefore just measures the mean square radius. 4. we can expand the exponential in (4. We can take over the result for electron-muon scattering and simply replace the mass of the muon by that of the proton: . This is because in the small |~q| limit the photon in Fig.1. the proton were a point charge e with Dirac magnetic moment e/2M. The above discussion cannot be applied directly to yield the structure of the proton. By virtue of the normalization condition (4.1. . (4. 6 where we have assumed that ρ is spherically symmetric. . . . If. Second. the proton’s magnetic moment is involved in the scattering of the electron. that is. yielding  Z  (~q . the proton is not static.2) F (0) = 1.5) = 1 − |~q| hr i + . ρ(r)r 2 d(cos θ) dφ dr 2 1 2 2 (4.1.3).1. but will recoil under the electron’s bombardment. . . not just its charge.~x)2 . and θ is the angle through which the electron is scattered. If |~q| is not too large. v = k/E. First. however.

 ′   E α2 q2 dσ .

.

(4. 2 θ 2 θ = cos − .1.7) sin dΩ .

9).2) is   Z 1 (4. q 103 . Copying the calculation of the electron muon cross section.8) given by (E. the lowest order amplitude for electron proton elastic scattering (shown in Fig.0.1.1. 4.lab 2 2M 2 2 4E 2 sin4 (θ/2) E where the factor E′ = E −1  2E 2 θ 1+ sin M 2 (4. arises from the recoil of the target.9) Tf i = −i ejµ − 2 (−e)J µ d4 x .

12) 2M where F1 and F2 are two independent form factors and κ is the anomalous magnetic moment. p′ . it does not make any difference that the proton has structure at order of 1 fermi. and the Dirac γ matrices.11) Since the proton is an extended structure. q.11) by γ µ .7). respectively ′ (4. We effectively see a particle of charge e and magnetic moment (1+κ)e/2M. and experimentally κn = −1. where κ. (Terms involving γ 5 are ruled out by conservation of parity. that is.10). where q = p − p′ and the electron and proton transition currents are. The corresponding values for the neutron are F1 (0) = 0.1.21 particles in (4. we know that J µ must be a Lorentz four vector.1.1.x . as for point spin. If we use (4. i   h κ 2 µ 2 µν = F1 (q )γ + F2 (q ) iσ qν (4. and so we must use the most general four-vector that can be constructed from p.1.2: Lowest-order electron-proton elastic scattering.12) to calculate the differential cross section for electronproton elastic scattering.) For q 2 → 0.1.1. the anomalous moment.1. is measured to be 1. F2 (0) = 1.k k′ p p′ Figure 4. .91. x . The factors in (4. when we probe with long-wavelength photons. we cannot replace the square brackets in (4.79.1. we find an expression similar to (4.12) must therefore be chosen so that in this limit F1 (0) = 1 and F2 (0) = 1. However. (4.10) ej µ = eu(k ′ ) γ µ u(k)ei(k −k) .   ′ −eJ µ = −eu(p′ ) u(p)ei(p −p) .

   ′  α2 κ2 q 2 2 E dσ .

.

2 = F1 − F cos2 (θ/2) dΩ .

1.13) 2M 2 104 .lab 4M 2 2 4E 2 sin4 (θ/2) E  q2 2 2 − (F1 + κF2 ) sin (θ/2) . (4.

kown as the Rosenbluth formula. it is better to use linear combinations of the F1.13) would revert to (4. These form factors can be determined experimentally by measuring dσ/dΩ as a function of θ and q 2 .1. F1.14) defined so that no interference terms.2. GE GM .13) then becomes .3 The two form factors. 4.2 GE ≡ F1 + κq 2 F2 .1. Note that if the proton were point-like like the muon.2 (q 2 ).1. occur in the cross section. 4M 2 GM ≡ F1 + κF2 . then κ = 0 and F1 (q 2 ) = 1 for all q 2 .7).1. (4. and (4. In practice. parametrize our ignorance of the detailed structure of the proton represented by the blob in Fig. Equation (4.

   ′ 2 α2 dσ .

.

cos + 2τ GM sin dΩ . E GE + τ G2M 2 θ 2 2 θ = .

Now that interference terms have disappeared. Rosenbluth. Having measured the size of the proton. Rev. The result for GE (q 2 ) is  −2 q2 GE (q ) ≃ 1 − 0.lab 1+τ 2 2 4E 2 sin4 (θ/2) E (4. respectively.1. The data on angular dependence of ep → ep scattering can be used to separate GE . Phys. (4. In particular. one might like to take a more detailed look at its structure by increasing the −q 2 of the photon to give 3 M.16) The behavior for small −q 2 can be used to determine the residual terms in the expansion of (4.1.8 fm is obtained for the magnetic moment distribution.1). GE and GM are referred to as the electric and magnetic form factors. GM at different values of q 2 . N. these proton form factors may be regarded as generalizations of the non-relativistic form factor introduced in (4.1.15) 2 2 with τ ≡ −q /4M . the mean square proton charge radius is   dGE (q 2 ) 2 = (0.71 2 (in units of GeV2 ). 105 . (4.81 × 10−13 cm)2 .1.5). 615 (1950).1.17) hr i = 6 dq 2 2 q =0 The same radius of about 0. 79.

In this case.3.2.2 would therefore need to be generalized to Fig. 4. Therefore.2. one might just excite the proton into a ∆-state and hence produce an extra π-meson. The square of the invariant amplitude (3. The hadronic tensor W µν serves to parametrize our ignorance of the form of the current at the end of the propagator. This is inadequate to describe the inelastic events of Fig. The most general form of the tensor W µν must now be constructed out of g µν and the independent momenta p and q (with p′ = p+q). that is ep → e∆+ → epπ 0 .18) where Lµν represents the lepton tensor of (3.11). J µ must have a more complex structure than (4.k′ k hadrons proton Figure 4. however. and the most general form of Γµ was constructed.3 is left unchanged. µν W (e) (4.1. a catch: because of the large transfer of energy. since everything in the leptonic part of the diagram above the photon propagator in Fig. The switch from a muon to a proton target was made by replacing the lepton current j µ (∼ uγ µ u) by a proton current J µ (∼ uΓµ u).2. The picture of Fig.41). 4. 4. the proton will often break up.1. the debris becomes so messy that the initial state proton loses its identity completely and a new formalism must be devised to extract information from the measurement. however.10). 4. When −q 2 is very large.3: Lowest-order diagram for ep → eX.3 because the final state is not a single fermion described by a Dirac u entry in the matrix current.37) is generalized to µν |M|2 ∝ L(e) . the square of the invariant mass is W 2 ≃ M∆2 . γ µ is not included as we are parametrizing 106 . better spatial resolution. (4. 4.1. and Fig. For modest −q 2 . This can be done simply by requiring a large energy loss of the bombarding electron.11).1. There is. The problem facing us now is illustrated by recalling (4.

1. so that the virtual photon probe is replaced by a weak boson.1. q) + 2 q 2 = 0 2 M M which lead to W4 = W5 = − p.1.|M|2 which is already summed and averaged over spins.23) (4. M2 M M (4.20) Setting the coefficients of q µ and pµ to zero.18) because the (e) tensor Lµν is symmetric.q ν qµqν µ ν µν µν + W2 2 p − 2 q p − 2 q .q q2 2 M2 W1 .q W2 . Note the omission of W3 in our notation. q)q µ]. q)pµ + 2 q 2 q µ + 2 [q 2 pµ + (p .21) (4. this spot is reserved for a parity violating structure function when a neutrino beam is substituted for the electron beam.1.22) (4. we find −W1 + W4 2 W5 q + 2 (p .1. We write W µν = −W1 g µν + W2 µ ν W4 µ ν W5 µ ν p p + 2 q q + 2 (p q + q µ pν ) .q µ p. q) = 0 .19) We have omitted antisymmetric contributions to W µν .19) are independent. The current conservation at the vertex requires qµ W µν = qν W µν = 0.24) Hence. q2 (4.1. W = W1 −g + 2 q M q q (4. and we can write without loss of generality      1 p.1. 0 = qν W µν = −qν W1 g µν + W4 W5 W2 (p .1. since their contribution to the cross section vanishes after insertion into (4.1.25) 107 . only two of the four inelastic structure functions of (4. M2 M W2 W5 (p . q2  W2 + p. consequently. 2 M M M (4.

1. k ′ ] . and we choose q2 and ν ≡ p. 2E ′ (2π)3 (4. for Lµν . dσ = 1 4 [(k .28) In the laboratory frame.37)].1.3). p)2 − m2 M 2 ]1/2  e4 µν L Wµν 4πM q 4 (e)  d3 k ′ .29) see (E. k ) + 2W2 [2(p .0.where the Wi ’s are functions of the Lorentz scalar variables that can be constructed from the four-momenta at the hadronic vertex.25). k)(p . The extra factor of 4πM arises because we have adopted the standard convention 108 .2.1. Using the expression (3. (4.1.q . ep → eX. (4. M2 (4. M (4. q 2 ) sin2 2 2 2 2  . we find ′ Lµν (e) Wµν = 4W1 (k .2. By including the flux factor and the phase space factor for the outgoing electron.1. we can obtain the inclusive differential cross section for inelastic electron-proton scattering. (µ) given by (4. q ) cos + 2 W1 (ν.27) Evaluation of the cross section for ep → eX is straightforward repetition of the calculation for e− µ− → e− µ− scattering with the substituttion of Wµν .30) where |M|2 is given by the expression in the braces [recall (3. k ′) − M 2 k . this becomes Lµν (e) Wµν = 4EE ′  θ θ W2 (ν.1. Unlike elastic scattering there are two independent variables.42) for Lµν (e) and noting (e) (e) q µ Lµν = q ν Lµν = 0.26) The invariant mass W of the final hadronic system is related to ν and q 2 by W 2 = (p + q)2 = M 2 + 2Mν + q 2 .

1.for the normalization of W µν .29) in (4.1.30) yields . Inserting (4.

1 E ′ |M|2 dσ .

.

= dE ′ dΩ .

to obtain the final result we used (E.31) where we neglect the mass of the electron. = W2 (ν. It is often more convenient to express the differential cross section with respect to the invariants ν and Q2 .3).lab 16π 2 E 4πM (4πα)2 E ′ µν = L Wµν 16π 2 q 4 E   4α2 E ′ 2 2 2 θ 2 2 θ W2 (ν. q ) cos + 2W1 (ν. q ) sin = q4 2 2   2 α 2 2 θ 2 2 θ . q ) cos + 2W1 (ν.1.0. q ) sin 2 2 4E 2 sin4 (θ/2) (4.

.

π dσ .

.

dσ .

.

= dQ2 dν .

lab EE ′ dE ′ dΩ .

E ′ . Q4 E 2 2 In experimental settings one may mantain the same values of Q2 and ν upon changing E. ν) sin . and then in principle could separate the two structure functions W1 and W2 . the differential cross section in the energy (E ′ ) and angle (θ) of the scattered electron can be written in the form . it is useful to make a compendium of our results on form factors.lab   4πα2 E ′ 2 2 θ 2 2 θ = W2 (Q . For all the interactions. and θ. We keep to the laboratory kinematics (see Appendix E) and neglect the mass of the electron. ν) cos + 2W1 (Q . For future reference.

  4α2 E ′2 dσ .

.

32) dE ′ dΩ . = . (4.1.

      q2 q2 2 θ 2 θ δ ν+ . for a muon target of mass m (or a quark target of mass m after substitutions α2 → α2 e2q where eq is the quark’s fractional charge).1.lab q4 First. (4.33) = cos − sin 2 2m2 2 2m eµ→eµ 109 .

1.34) cos + 2τ GM sin 1+τ 2 2 2M ep→ep where τ = −q 2 /4M 2 and M is the mass of the proton.34) can be integrated over E ′ with the result [see (E. q 2 ) cos2 + 2W1 (ν. q 2 ) sin2 .0.1.1.12)] . (4.For elastic scattering from a proton target    2    q2 GE + τ G2M 2 θ 2 2 θ = δ ν+ (4.33) and (4. (4. Finally.35) 2 2 ep→eX Making use of the delta function.1. for the case when the proton target is broken up by the bombarding electron   θ θ = W2 (ν.

  dσ .

.

36) = dΩ .1. (4. E′ α2 .

(4. (The “point” notation reminds us that the quark is a structureless Dirac particle. we should be able to illuminate them with a small wavelength (large −q 2 ) virtual photon beam. The sign that there are structureless particles inside a complex system like a proton is that for small wavelengths.1. the proton described by (4. point-like.38) 2mν 110 .35) suddenly starts behaving like a free Dirac particle (a quark) and (4. Q ) = .1.1.) Using the identity δ(x/a) = aδ(x).1.37) may be rearranged to introduce dimensionless structure functions   Q2 Q2 point 2 2mW1 (ν.35) turns into (4.37) δ ν− 2W1 = 2m2 2m 2m where Q2 ≡ −q 2 and m is the quark mass.33).1.lab 4E 2 sin4 (θ/2) E If simple. (4.21 quarks reside inside the proton. The proton structure functions thus become simply     Q2 Q2 Q2 point point W2 =δ ν− . Q ) = δ 1 − .1. spin. The fact that such photons break up the proton target can be handled by using the inelastic form factors described above. δ 1− 2mν 2mν   Q2 point 2 νW2 (ν. (4.

D.4 The mass m merely serves as a scale for the momenta Q2 . The point structure functions.71 GeV)2 . 1547 (1969).1. p)/Q2 = 2Mν/Q2 . (4. the structure functions of (4.1. F2 (ω) .40) Note that in (4. In the late 60s.1.39) contain a form factor G(Q2 ). the structure functions were approximately independent of Q2 . and so cannot be rearranged to be functions of a single dimensionless variable. 179. (4. The data exhibited Bjorken scaling to about 10% accuracy for values of Q2 above 111 . on the other hand.37). A mass scale is explicitly present. Rev. For simplicity. set κ = 0. We have introduced the proton mass instead of the quark mass to define the dimensionless variable ω. Q2 ) 7−→ large Q2 7−→ large Q2 F1 (ω) .39) W2elastic = G2 (Q2 ) δ ν − 2M In contrast to (4. the proton is more likely to break up. Bjorken. it is set by the empirical value 0. The presence of free quarks is signaled by the fact that the inelastic structure functions are independent of Q2 at given value of ω.1.These “point” functions now display the intriguing property that they are only functions of the ratio Q2 /2mν and not Q2 and ν independently. then comparing (4.1. As Q2 increases above (0.1. and no scale of mass is present.1. The so-called “Bjorken scaling” can be summarized as follows: in the limit Q → ∞ and 2Mν → ∞ such that ω = 2(q . deep inelastic scattering experiments conducted by the SLAC-MIT Collaboration showed that at sufficiently large Q2 ≫ Λ2QCD . so that GE = GM ≡ G.34) and (4. ν.40) we have changed the scale from what it was in (4.1.38). Q2 ) νW2 (ν.35) we have   Q2 2 2 Q2 elastic W1 = G (Q ) δ ν − 4M 2 2M   2 Q . the structure functions would have the following property MW1 (ν. This behavior can be contrasted with that for ep elastic scattering. the form factor depresses the chance of elastic scattering. depend only on a dimensionless variable Q2 /2mν. Phys.5 4 5 J.71 GeV in the dipole formula for G(Q2 ) which reflects the inverse size of the proton.

of partons) within the proton is slowed down by time dilation. 4. Rev. 4. Paschos. M. 23. 1975 (1969). the proton is Lorentz-contracted into a thin pancake. P. S. its energy can only be xE if we put m = M = 0. is to represent the inelastic scattering as quasi-free scattering from point-like constituents within the proton. Feynman.4: Kinematics of lepton-proton scattering in the parton model. Clearly.6 The basic idea in the model. Phys. when viewed from a frame in which the proton has infinite momentum. 23. 32. Breidenbach et al. E. Lett. Rev. Equivalently. Lett. Ann. 203 (1972).” In this frame. shown in Fig. Nucl. 6 R. I. J. Moreover. the proper motion of the constituents (i. Bjorken and E. Kendall. 935 (1969). We envisage the proton momentum p as being made of partons carrying longitudinal momentum pi = xi p. 23. D.. A. Lett. D. Phys. Friedman and H.41) partons (i) Assigning a variable mass xM to the parton is of course out of the question. Part. p~ ≫ M the so-called “infinite momentum frame. if the parton’s momentum is xp. Poucher et al..2. Bloom et al. where the momentum fractions xi satisfy: X 0 ≤ xi ≤ 1 and xi = 1 .4.2 Parton Model Now that scaling is an approximate experimental fact. Rev..e. Rev. i.e. 1415 (1969). Phys. we adopt the spirit of the parton model. 930 (1969). Phys. 185. (4. Furthermore. Rev.k′ k q xp X p (1 − x)p Figure 4. Phys. J. Imagine a reference frame in which the target proton has a very large 3-momentum. J. W. Rev. 22. and the lepton scatters instantaneously. Lett.. 112 . because of the large momentum transfer (−q 2 ≫ M) interactions between partons can be neglected and therefore the individual current-parton interactions may be (1 GeV)2 . 118 (1974). Sci. a proton can only emit a parton moving parallel to it (p⊥ = 0 for both) if they both have zero mass.

treated incoherently .

dσ .

.

= dtdu .

ep→eX X partons(i) Z .

dσ .

.

dxfi (x) . dtdu .

2.3.44). With this in mind. q 2Mν (4.2. from (4.44) we have x(s + u) + t = 0.46) can be rewritten as .45) Inserting (4. tˆ2 (4. therefore Consequently. and the the sum is over all the contributing partons.48) become sˆ = (k + xp)2 ≃ x(2k .2. (4.45) into (3. p) ≃ xs .eqi→eqi (4.2.2.44) ′ 2 ′ uˆ = (k − xp) ≃ x(−2k . |M|2 = 2 (4παeq )2 sˆ2 + uˆ2 . or sˆ + uˆ + tˆ = 0. tˆ = (k − k ′ )2 = t = q 2 .46) sˆ2 dtˆ tˆ2 Using the invariant relations (4. − s+u 2p .2.2. the invariant variables of (3. Assuming s ≫ M.3. p) ≃ xu .2.27) we obtain an expression for the differential cross section   2πα2 e2q sˆ2 + uˆ2 dσ = . (4.1.43) q2 Q2 t =− = = x. Eq. the invariant amplitude follows directly from (3. (4.42) where fi (x) indicates the probability of finding constituent i inside the proton.48).2.

dσ dσ .

.

= x .

dtdu eqi →eqi
dtˆdˆ
u  

Z
2πα2 e2q sˆ2 + uˆ2
d
= x
δ(ˆ
s + uˆ + tˆ)dˆ
u

u
sˆ2
tˆ2  

2πα2 e2q s2 + u2
δ(x(s + u) + t)) .
(4.2.47)
= x
t2
s2
Now, we can rewrite (4.1.28) in terms of the invariant variables
Lµν
(e) Wµν = −2W1 t + 

W2 
2
−su
+
M
t
,
M2

113

(4.2.48)

and because we assume s ≫ M 2 , we have
Lµν
(e) Wµν =

2
[x(s + u)2F1 − suF2 ] ,
M(s + u)

(4.2.49)

where t = −x(s + u), F1 ≡ MW1 and F2 ≡ νW2 . Substituting (4.2.49) into
(4.1.31) we have


4πα2 1 

50) 1 2 dtdu . (4. 2 = (s + u) xF − usF .2.

2. here fixed by the delta function in (4.2 (x).42) and comparing coefficients of us and s2 + u2 .47) and (4.50) into (4.2.2. (4. we can redefine F1.2.3).2. and F1 (ω) = (ω/2)F2 (ω). and t = −Q2 = −4EE ′ sin2 (θ/2)   1 4πM 2 dt − du . dΩdE = 2πd(cos θ)dE = su 2M ′ ′ (4.k E (4.2.ep→eX t2 s2 s + u where we have used the kinematic relations in the lab frame (see Appendix E) u = −2ME ′ .47).51) (4. using the lab frame kinematic relation (E.0. s = 2ME.2 (ω) as F1. νW2 (ν. we obtain sin2 and Q2 2Mνx xyM θ = = = ′ ′ 2 4EE 4E ν/y 2E ′ E θ = ′ cos 2 E 2 where y=   Mxy . |{z} p.40). P namely. Recalling the identification (4.53) i We see that F1 and F2 are functions only of the scaling variable x. 1−y− 2E p. Next.54) (4.1.q ν = .52) Substituting (4. we see that.56) (lab) P R Note that F2 (ω) = i dxe2qi fi (x) x δ(x − 1/ω). Q2 ) 7→ F2 (x) = i e2qi xfi (x) and M W1 (νQ2 ) 7→ F1 (x) = F2 (x)/(2x).2.55) (4.2.2. we obtain the master formula of the parton model7 X 2xF1 (x) = F2 (x) = e2qi x fi (x) . at large Q2 .2. 7 114 .

though the dependence on the latter is trivial.8 This relation gave evidence that the partons involved in deep inelastic scattering were fermions.2. we obtain the Callan-Gross relation X dσ 2πα2 2 = s [1 + (1 − y) ] e2qi x fi (x) . L. The relation between Q2 and θ follows from (E.2.2. Gross. Manohar. whereas lines of fixed x remain constant. (eds.2. Lett. (4. 4 dxdy Q i (E ≫ Mx).5. lines of constant θ are straight lines passing through the point ν/E = 1. There are three independent variables which describe the kinematics: E ′ .60) Therefore. 9 115 .59) is specific to the scattering of electrons from massless fermions. at a time when the relation between partons and quarks was still unclear.2.57) where we have used the identity dE ′ dΩ = π 2ME dQ2 dν = π y dx dy . 0 ≤ ν ≤ E. Kamal. Singapore. 22. 0 ≤ x ≤ 1. .9 The boundary of the physical region is given by the requirements that 0 ≤ θ ≤ π.55) into (4. ′ EE E′ (4. 4. Callan and D. as shown in Fig. A.54) and (4. Lines of fixed θ become steeper as the beam energy increases. Greeniaus.53) into (4.2.Substituting (4.1. World Scientific. in Symmetry and Spin in the Standard Model. θ. Khanna. Phys. G. C. J.1. A. and intersecting the Q2 /2ME axis at Q2 /2ME = (E/M)(1−cos θ). The Q2 dependence of the kinematic variables is crucial to understand which terms are important in the deep inelastic 8 C.p.57). G. Because x = Q2 /2Mν = (Q2 /2ME)/(ν/E). the contours of constant x are straight lines through the origin with slope x. and φ.59) The behavior [1 + (1 − y)2 ] in (4. N.2.2. 156 (1969). V. B. Rev. F. 1992).58) Substituting (4.31) we get     8MEπα2 Mxy dσ 2 F2 = xy F1 + 1 − y − dxdy Q4 2E (4.2.3) and is given by 1 Q2 = (E − ν)(1 − cos θ). Campbell. It is convenient to plot the allowed kinematic region in the (Q2 /2ME) − (ν/E) plane.0. A. 2ME M (4.

Lett. Therefore. (E ≫ Mx). 412 (1993). Abt et al. It started operating at the end of 1991 and ceased running in June 2007. I. B 321.5 that for fixed beam energy E. Phys. A generic point in the kinematic plane is given by some value of x and y. Two experiments.2. the maximum allowed value of Q2 is at the intersection of the line θ = π with the line for fixed x. The small x region is also the small Q2 region.61) Qmax = 2MEx 2E + Mx To be in the deep inelastic region. [ZEUS Collaboration]. [H1 Collaboration]. and assuming Q2 ≥ 10 (GeV)2 is large enough to be considered deep inelastic scattering. 4. pQCD seems to give a very good description of the F2 behaviour down to low values of momentum transfers squared: Q2 of 10 M. We can also see in Fig. For a fixed value of x. H1 and ZEUS. Phys. because lines of constant x approach the horizontal axis for small x. there is a limit to the Q2 −x region which can be explored experimentally. For example.limit. for a fixed value of x and y. B 407. This corresponds to s = 4 × 28 × 820 (920) (GeV)2 . 116 . This implies that a generic point in the physical region has (1−cos θ) ∝ M/E ∝ M 2 /Q2 . θ ∝ M/Q. Lett. the variables ν/E and Q2 /2ME are fixed. 515 (1993). the smallest measurable value of x is 10−2 . so this places a limit on the smallest value of x accessible for a given beam energy. one needs Q2 to be larger than a few (GeV)2 . It is elementary to find the intersection point of the two lines. 161 (1994). As E → ∞.5 GeV and protons accelerated to an energy of 820 GeV until 1997 and 920 GeV starting from 1998 onwards. (A similar measurement in a fixed target experiment would require a 50 TeV lepton beam. in the deep inelastic regime ν ∝ Q2 /M and E ∝ Q2 /M. Phys. collected data from collisions of e− or e+ with an energy of 27. goes to zero as Q2 → ∞.) One of the first important results of the H1 and ZEUS measurements was the observation of a steep rise of the proton structure function F2 towards low values of the Bjorken variable x. The Hadron-Elektron-Ring-Anlage (HERA) at DESY was the first ever constructed storage ring to collide positrons or electrons with protons. B 316. and hence the scattering angle. (4. Nucl. allowing measurements of structure functions down to x ≈ 10−4 .   2E 2 ≈ 2MEx. Derrick et al. with a 500 GeV lepton beam. Furthermore.10 This phenomenon has been successfully described by (perturbative) pQCD calculations.

In the deep inelastic limit. 4. The dot-dashed lines are curves of constant scattering angle θ.3 QCD Improved Parton Model The simple parton model described in the previous section is not true in QCD. much of the structure of the parton model remains in perturbation theory.5: The triangle is the allowed kinematic region for deep inelastic scattering.Q2 ___ 2ME __ 2E M θ=π ∞ __ E M x=1 _ θ= π 2 x=_21 2 Qmax for x=_21 v/E 0 1 Figure 4. The former contains only high energy and momentum components and. because the properties we assumed for the hadronic blob are explicitly violated by certain classes of graphs in perturbation theory. The latter piece describes non-perturbative physics. Factorization permits scattering amplitudes with incoming high energy hadrons to be written as a product of a hard scattering piece and a remainder factor which contains the physics of low energy and momenta. the order of a few GeV2 . Nevertheless. because of asymptotic freedom. The dashed lines are lines of constant x. is calculable in perturbation theory. because of the property of factorization. the intercept of the constant θ lines with the vertical axis → ∞. but is described 117 . We discuss this next.

Soper and G. Collins. G. αs (µ2 )) fi (x. the sum i runs over all partons in the incoming hadron. D. but it is assumed to have a validity which transcends perturbation theory. 118 . Phys. K. 263 (1984). Nucl.62) depend on the renormalization and factorization scales. 12 Indeed all quantities in (4. The structure functions themselves are not calculable in perturbation theory. Politzer and G.3.62) i 0 where µ2 is the large momentum scale which characterizes the hardness of the interaction. J. and σ ˆ is the short distance cross section calculable as a perturbation series in the QCD coupling αs . B 134. 285 (1979). Assuming the property of factorization holds we can derive the QCD improved parton model. In order to perform the factorization we have introduced a scale µ2 which separates the high and low momentum physics. p) = dx σ ˆ (|q|. The presence of infrared singularities or singularities coming from regions of collinear emission reveals the sensitivity of a Feynman graph to very low momentum scales. B 152. xp. such singularities are associated with real physical processes rather than virtual processes which occur only as short-live fluctuations. Phys. Sterman.11 The plausibility of the factorization property can be seen from the following argument. it is appropriate that they are included in the wave function of the incoming hadron and not in the short distance cross section. H. Because these real processes occur long before the hard interaction. H. µ2 ) .by a single process independent function for each type of parton called the parton distribution function (PDF). Ellis. M. The result for any process with a single incoming hadron leg is XZ 1 σ(|q|. Because of the Landau rules. Georgi. D. The proofs of factorization establish that this simple picture is in fact valid in perturbation theory. C. The factorization has been proven within perturbation theory. It is referred to as the short distance cross section because the singularities corresponding to a long distance physics have been factored out and abosorbed in the structure functions fi . E. The proofs require a detailed examination of all the dangerous regions of phase space in Feynman graphs. Ross. Without this property of factorization we would be unable to make predictions for processes involving hadrons using perturbation theory. (4. Machacek.3. Lett.12 No physical results can depend on the particular 11 R.

781 (1972) [Sov. which is the standard approximation.3.67) = i qq qg d ln µ2 2π x z which are usually taken to be the same (µr = µf = µ). Nucl. Sov. 641 (1977) [Zh. Altarelli and G. 298 (1977).3. Lipatov. Fiz. d ln µ2 (4. N. J. Teor. fi (x. although often still with large uncertainties. 13 V.6. JETP 46. Yad. J. G. 675 (1972)]. 15. Gribov and L. Phys.3. . (4. µ2) x z j ! qj (x/z. L. This implies that any dependence on µ in σ has to vanish at least to order in αs considered. µ )P (z) + g(x/z. Nucl. Y.64) where the matrix P is calculable as a perturbation series αs (1) (0) Pij (z. d σ (n) = O(αsn+1 ) .65) are needed for nextto-leading order (NLO) predictions. 73. µ2 ) αs (µ2 ) 1 dz  2 2 q (x/z. the splitting functions Pij are known to NNLO. . 1218 (1972) [Sov.value chosen for this scale.63) The evolution of the parton distributions with changes of the scale µ are predicted by the Dokshitzer-Gribov-Lipatov-Altarelli-Parisi (DGLAP) equation. Currently. (4. N.3. Dokshitzer. Eksp. µ2) . Phys. Parisi. 119 . Phys. 15. B 126. αs (µ2 )) fj (ζ. Phys.13 Z αs (µ2 ) X 1 d 2 dz dζ δ(x − zζ) Pij (z. (4. 438 (1972)]. 1216 (1977)]. Nucl. µ2 ) which is a system of coupled integro-differential equations corresponding to the different possible parton splittings Z  dqi (x.3. Fiz.3. µ )P (z) . Yad. The first two terms of (4.65) 2π ij Examples of Feynman diagrams contributing to P in leading order QCD are shown in Fig. 15. αs ) = Pij (z) + P (z) + . Performing the ζ integration we obtain ! ! Z Pqi qj (z) Pqi g (z) qi (x. µ2 ) × . 15. Fiz. µ2 ) αs (µ2 ) X 1 dz d = d ln µ2 2π Pgqj (z) Pgg (z) g(x.66) g(x/z. 4. µ ) = d ln µ2 2π x j (4. .

µ2 )Pgq (z) + g(x/z. a single parton of transverse dimension 1/µ is resolved into a greater number of partons.p zp p Pqq(z) zp Pqg(z) p zp p Pgq(z) zp Pgg(z) Figure 4. αs (µ2 ) X dg(x. µ ) again relies on the infinite momentum frame. µ2 )Pgg (z) . qq qq gq 0 0 0 where nf is the number of flavors. These equations correspond to quark number conservation and momentum conservation in the splittings of quarks and gluons. dz x[P (z) + P (z)] = 0. The interpretation as probabilities implies that the DGLAP are positive Rkernels R 1 definite for z < 1. In this frame fj (x. Viewed on a smaller scale of transverse dimension r ′ . They satisfy R 1 the following relations: 1 dz P (z) = 0. 120 . We indicate the collinear momentum flow (p incoming and zp outgoing) as it enters the calculation of the corresponding splitting function Pij . such that r ′ ≪ 1/µ.3. As we increase µ.6: Sample of Feynman diagrams for parton-parton splitting in leading order QCD. the DGLAP equation predicts that the number of partons will increase. 2 The DGLAP kernels Pij have an attractive physical interpretation as the probability of finding parton i in a parton of type j with a fraction z of the longitudinal momentum of the parent parton and transverse size less than 1/µ. µ2 ) is the number of partons of type j carrying a fraction x of the longitudinal momentum of the incoming hadron and having a transverse dimension r < 1/µ. µ2 ) = d ln µ2 2π j Z x 1  dz  qj (x/z. z (4.68) The physical interpretation of the PDFs fj (x. and dz z[2 nf Pqg + Pgg ] = 0.

[ZEUS Collaboration]. 4.69) 4 1 + (1 − z)2 . J. 012 (2002). D. 189 (2009). Eur. Huston. M. C 63. L.72) 1−z z In the double-leading-logarithmic approximation. 4. 1 (2003).2. The gluon radiation explains the F2 scaling violation. Martin. 15 In the framework of QCD a proton consists of three valence quarks interacting via gluon exchange. as displayed in Fig.68) to be sufficiently accurate in pQCD. an input distribution at Q20 = 10 GeV2 can be determined in a global fit from comparison to HERA data. The DGLAP evolution equations (4. i. Nadolsky and W.3.8.3.15 The large difference in the hard squared momentum scale Q2 between HERA and LHC requires the parton evolution based on Eqs. W. Pumplin. shown in Fig. The gluons can produce virtual quark-antiquark pairs. 4. H.3. that is limx→0 ln(1/x) and limQ2 →∞ ln(Q2 /ΛQCD ). J. Benchmark CTEQ and MSTW parametrizations from global fits of hard-scattering data account for the effects of experimental errors and come with the according uncertainties. A. 16 J. J. Pqq (z) = 3 1−z (4. in agreement with the experimental results from HERA. however. Chekanov et al. Tung. JHEP 0207. J. and. R. Stirling. 121 . C.41) and (2.2. Thorne and G.e.3. Eur. For example.14 The PDFs. Pgq (z) = 14 S. R.. Phys.67) and (4.3. [H1 Collaboration].42). cannot be calculated “from first principles” in pQCD.67) and (4. (4.3. K.71) Pqg (z) = 2 and   z 1−z Pgg (z) = 6 + + z(1 − z) .9. Phys. P.68) are solved by inserting certain analytical functions at some starting scale Q20 and evolving them up to higher Q2 .16 An example is given in Fig. Phys. (4. the DGLAP equation predicts a steeply rising gluon density at low x. D 67. Rev. (4. Lai. because of their selfcoupling (2. which shows the NLO PDFs at scales of Q2 = 10 GeV2 and Q2 = 104 GeV2 .3. The structure function F2 found as a result of this procedure is adjusted to the experimentally measured one. (4. D. S. Adloff et al. 012007 (2003). other gluons.70) 3 z z 2 + (1 − z)2 .7. Watt.The DGLAP kernels at LO become 4 1 + z2 .3. C 30. the F2 dependence on Q2 . Stump. including the associated 68%CL uncertainty bands. so-called sea quarks.

10 7.7: Gluon momentum distributions xf (x.2 xg(x. Q2=200 GeV2 15 12.Q ) 2 2 Q =20 GeV 20 H1 NLO-QCD Fit 2000 17.5 0 -4 10 10 -3 10 -2 10 -1 X Figure 4. uncert. 122 .) 2001 exp. exp.5 ZEUS NLO-QCD Fit (Prel. uncert. Q2 ) in the proton as measured by the ZEUS and H1 experiments at various Q2 .5 5 2 Q =5 GeV 2 2.5 total uncert.

2 Q2 = 10 GeV2 1 1.6 MSTW08 68% CL 0.2 0 10-4 10-3 10-2 10-1 0 10-4 1 10-3 10-2 10-1 x 1 x Figure 4.s 10 -3 d c.2 0. 123 .Q2) xf(x.2 d 10-1 u 1 x 0 10-4 10 -3 10-2 d 10-1 1 x Figure 4.L.8 xg (× 0.c c. xf(x.) xuv 0. Q2 ) in the proton as measured by the ZEUS and H1 experiments at Q2 = 10 GeV2 are compared to the MSTW (left) and CTEQ (right) parametrizations.6 HERAPDF0.b 0.Q2) MSTW 2008 NLO PDFs (68% C.4 0.05) 0.1(prel.4 d 0 10-4 s.05) HERAPDF0.6 0.4 xuv CTEQ6.Q2 = 10 GeV2 1 xf xf 1 0.4 xdv xS (× 0.8: The valence.2 u 10-2 u 0.1(prel.5M 68%CL 0.) 1. Q2 ) in the proton from a low scale at Q2 = 10 GeV (left) to LHC energies at Q2 = 104 GeV (right).05) xdv xS (× 0.8 Q2 = 10 GeV2 0.6 b.2 Q2 = 104 GeV2 1 g/10 g/10 0.s s. sea and gluon momentum distributions xf (x.) xg (× 0.c 0.9: Evolution of gluon and quarks momentum distributions xf (x.05) 0.8 0.8 u 0.

This gives (3.In passing. and hence the coupling of a quark on a longitudinally polarized photon becomes possible.83). For more detailed calculations. x2 . µ2 ) ijk × σ ˆij→k (x1 . for e+ ǫ− → q q¯.73) where Dk→z (z. µ2 )Dk→X (z. To describe the fragmentation of quarks into hadrons. Like f functions. αs becomes large. Q2 .3. z. for a cross section σpp→X of some hadronic final state X in. helicity may not be conserved in this process. µ2 ) . this problem cannot be sidestepped. the produced quark and antiquark separate with equal and opposite momentum in the center-of-mass frame and materialize into back-to-back jets of hadrons which have momenta roughly collinear with the original q and q¯ directions. For example. Eventually.5. µ2 ) is the fragmentation function and all other functions have a clear interpretation. (4. the D functions are subject to constraints imposed by momentum and probability con124 . we use an analogous formalism to that introduced to describe the quarks inside hadrons. and strong color forces pull on the separating q and q¯. µ2 ) fj (x2 . Namely. say. proton-proton scattering we can write σpp→X = XZ dx1 dx2 dz fi (x1 . The fragmentation function D(z). we have not faced the problem of how the quarks turn into hadrons that hit the detector. all the energy is degraded into two jets of hadrons moving more or less in the direction of the q and q¯. αs (µ2 ). The hadrons may be misaligned by a momentum transverse to the q or q¯ direction by an amount not exceeding about 300 MeV. The potential energy becomes so large that one or more q q¯ pairs are created. We can visualize jet formation as hadron bremsstrahlung once the q and q¯ separate by a distance of around 1 fm. It was sufficient to state that quarks must fragment into hadrons with unit probability. we should note that because of gluon exchange corrections in pQCD the longitudinal structure function FL could differ from zero: because quarks can have a non-negligible virtuality before scattering on the probing photon. describes the transition (parton → hadron) in the same way that the structure function f (x) describes the embedding (hadron → parton). So far. Thus.

Equation (4.75) 0 1 zmin where zmin is the threshold energy 2mh /Q for producing a hadron of mas mh . which is consistent with the so-called “leading-log QCD” behavior and seems to reproduce quite well the multiplicity growth as seen in colliders experiments. (4. in what follows we illustrate by two brief examples that hard scattering events 17 C. namely. (4. the average multiplicity per jet is approximately 54. Clearly the same relation holds for Dqh¯ (z). Equation (4. T. where nh is the average multiplicity of hadrons of type h.4 Physics of Hadronic Jets Jet studies in hadron-hadron collisions have traditionally been viewed as less incisive than those carried out in electron-positron annihilations or in lepton nucleon scattering because of the added complexity of events.3.3.75) says that the number nh of hadrons of type h is given by the sum of probabilities of obtaining h from all possible parents.08 exp 2. from q to q¯ of any flavor. B 224.6 ln(1/z) (1 − z)2 z ln(1/z) .17 With the infrared cutoff set to z = 10−3 . 125 .servation: XZ h XZ q 1 zDqh (z)dz = 1 . The main features of the jet fragmentation process derived from dnh /dz ≈ (15/16) z −3/2 (1 − z)2 (which provides a reasonable parametrization of Eq. 469 (1983). E is the energy of any hadron in the jet.74) [Dqh (z) + Dqh¯ (z)]dz = nh . Nucl.3. 4. (4.3.76) where z ≡ E/Ejet . Hill.3.76) for 10−3 < z < 1) are summarized in Table 4. can be cast in the following form h p i h p i−1 dnh ≈ 0. dz (4.1. and Ejet is the total energy in the jet.74) simply states that the sum of the energies of all hadrons is the energy of the parent quark. However.3. A parametrization of the fragmentation spectrum. Phys.

Rev. 40. D 18. Phys.1 Hadroproduction of Direct Photons Hadronic reactions producing large-k⊥ direct photons provide remarkable tests of perturbative QCD. Lett. D 39. Farrar. 19 P.052 0. R.0350 0.0350 0.007 0. M.002 take on a much simpler aspect at high energies. Fontannaz. These two subprocesses may even be disentangled by taking cross section differences of the type σp¯p→γ+ jet − σpp→γ+ jet . Phys. F. 4. Rev. Phys.0047 R z2 (dnh /dz) dz z1 3 3 9 9 30 R z2 z1 z (dnh /dz) dz 0.1: Properties of jet hadronization.0047 0. therefore.10. at the LHC.182 0.018 0.18 Because of the point-like coupling of the photons to the quarks. Lett. Compton scattering becomes the dominant process contributing to the prompt photon production over most of the kinematical region.0100 0.167 0. Werlen.069 zequivalent 0. 337 (1977).Table 4. J. Halzen and D. Baier. 3378 (1978).0750 0.155 0. 1117 (1978). 126 .062 0. 3275 (1989). Phys.0750 0. z1 0.) 18 G.4. F.19 In this section we show that. the trigger photon represents the full jet. 4. Owens and M. no (nonperturbative) decay function enters into the prediction. (The quark distributions can be taken from deep-inelastic scattering. R. the reaction pp → γ + jet provides a quite sensitive probe of the gluon distribution. the valence-quark and gluon properties in the incident particles can then be studied separately. q¯g → q¯γ and the annihilation process q q¯ → gγ. Moreover.0010 z2 1. Scott.0100 0. Aurenche. B 67. shown in Fig. Rev. M. and that there is no major impediment to detailed analyses.546 0.0000 0. starting at leading order only two subprocesses are relevant: namely the QCD Compton process qg → qγ. Thus.

g γ k k′ p q p′ q q q p′ p g γ k′ k Compton q g p q¯ k q p′ p k′ k γ g p′ q¯ γ k′ annihilation Figure 4. The differential cross section for direct-photon production is .10: Leading order processes contributing to direct photon production.

.

XZ .

σ .

.

′ dσ .

′ dˆ 2E 3 ′ .

Q) 2E 3 ′ . Q) fj (xb . dxa dxb fi (xa .

Q is the momentum transfer. .4. Q) and fj (xb . and the sum is over the parton species: g. which results in the dominant contribution to the total cross section at the LHC. s. d. Corrections from the other two processes can be computed in a similar fashion. c. Q) are PDFs. In what follows we focus on gq → γq. b. dˆ σ /d3 k ′ |ij→γk is the cross section for scattering of partons of type i and j according to elementary QCD diagrams. (4.77) = . k ′ (E ′ ) is the photon momentum (energy). q = u. The hard parton-level cross section reads. d k pp→γX d k ij→γk ijk where xa and xb are the fraction of momenta of the parent hadrons carried by the partons which collide. fi (xa .

X (2π)4 1 σ .

.

′ 2 1 ′ dˆ = |M|2 δ[(k + p − k ) ] 2E 3 ′ .

q = k − k ′ . d k gq→γq (2π)6 2ˆ s 4 spins X 1 1 2 1 |M|2 .78) δ(2p .4. and −q 2 = −tˆ = Q2 . The 127 . q + q ) = (2π)2 2ˆ s 4 spins where k and p are the momenta of the incoming partons. (4. the parton-parton center-of-mass energy sˆ = xa xb s.

− s/2) we obtain √ √ s s ′ p2 . k0′ = k⊥ cosh y. Likewise.80) 4 9 sˆ + tˆ tˆ spins Equation (4. q + q 2 = 2 xb p2 . 0. 4 spins 3 sˆ sˆ + tˆ (4. (4. p1 = ( s/2. and eq is the fractional electric charge of species q.result   1X sˆ + tˆ 1 2 2 2 sˆ 2 + |M| = gs e eq .4. (4.81) ′ 2E 2 Considering that the incoming momentum of the gluon is k = xa p1 and that of the quark is p = xb p2 .   sˆ + tˆ 8 2 2 2 tˆ 1X 2 − |M| = gs e eq − . (4. s/2). (4.1 and insertion of the color factor 1/6 (see Appendix F).4.4.78) can be most conveniently integrated in terms of the rapidity y and transverse momentum k⊥ of the final photon 1 d3 k ′ = d2 k⊥ dy = πk⊥ dk⊥ dy .4.83) 2 2 and √ √ s ′ ˆ t = −2k .4. 0. for q q¯ → gγ. we can re-write the argument of the delta function as 2p .79) follows directly on substitution of α2 → e2q ααs in the corresponding QED amplitude given in Table 3. Intro√ √ ducing.4.85) xa − x⊥ ey 128 . k ′ + tˆ .4.84) 2 so that √ √ 1 δ(xa xb − xb x⊥ ey − xa x⊥ e−y ) δ(xa xb s − s xb k⊥ ey − s xa k⊥ e−y ) = s 1 = s [xa − x⊥ ey ]   xa x⊥ e−y × δ xb − . 0. and √ √ p2 = ( s/2.4. 0. kk′ = k⊥ sinh y. Recall that gs and e are the QCD and electromagnetic coupling constants. k = k⊥ (cosh y + sinh y) = k⊥ ey (4. (xa p1 − k ′ ) + tˆ = xa xb s − 2 xb p2 .82) where p1 and p2 are the initial momenta of the parent protons. k = −2xa k⊥ e−y = − s k⊥ e−y xa . (4.

4. Q) fq (xb .4.4. sˆ xb s xb z xa (4. (4. The upper bound xb < 1 leads to a stronger constraint x⊥ ey . (4.min "   #  −1 X x⊥ z xa 1 x x z a ⊥ 2 . (4. Putting all this together.88) becomes qg→γq σpp→γX Z 1/2 Z zmax Z 1 e2 gs2 = dx⊥ dz dxa fg (xa . Q) = 2 z q   1 xa x⊥ z −1 × δ xb − (2π)2 2xa xb s2 (xa − x⊥ z) xa − x⊥ z   e2 gs2 e2q sˆ + tˆ sˆ + × .89) Eq.4. Of course there is another completely symmetric term. the total contribution from gq → γq reads qg→γq σpp→γX Z X Z d3 k ′ Z 1 = 2 dxa dxb fg (xa .87) + × 3 sˆ sˆ + tˆ With the change of variables z = ey Eq.4.87) can be rewritten as qg→γq σpp→γX Now.86) which requires x⊥ ey < 1 − x⊥ e−y . Q) fq (xb .Q + × eq fq 2 x − x z x x x z a ⊥ a ⊥ a q 129 .4.4. in which g comes from p2 and q comes from p1 . Q) 12πs x⊥min zmin xa. yielding x⊥ < (2 cosh y)−1. since Z X Z π k⊥ dk⊥ dz Z dxa dxb fg (xa .(4. xa > 1 − x⊥ e−y (4. The lower bound xb > 0 implies xa > x⊥ ey . (4.90) .88) 3 sˆ sˆ + tˆ √ sk⊥ e−y x⊥ x⊥ z tˆ =− =− = −1. Q) ′ 2E (2π)2 q   1 xa x⊥ e−y 1 × δ xb − s [xa − x⊥ ey ] 2ˆ s xa − x⊥ ey   e2 gs2 e2q sˆ + tˆ sˆ .√ where x⊥ = k⊥ / s.

k⊥. |y| < 2.90). It is clearly seen that the gq → γq process provides the dominant contribution. as obtained through numerical integration of Eq. 130 .11 shows the leading order QCD cross section σpp→γ+jet vs k⊥. 100. the advantages of direct photons as a clean probe of parton distributions are offset by large QCD backgrounds which are about 102 to 20 L. Taylor. has been included in the calculation. (4.min. S.4. Bayatian et al.86).11: Leading order QCD σpp→γ+jet vs. 21 G. where the integration limits. R.4. 016005 (2008). Phys. Rev. D 78. Lett. J.√ Figure 4. for s = 14 TeV.91) 1 − x⊥ z −1 are obtained from Eq. (4.21 Unfortunately. Phys. A. Nawata and T. Figure 4. [CMS Collaboration]. 995 (2007).min.4. H.4. Anchordoqui.min = x⊥ z . L. an additional kinematic cut. G 34. 171603 (2008).20 To accommodate the minimal acceptance cuts on final state photons from the LHC experiments. Goldberg. Phys. (4. Rev. s " # 1 1 1 ± −4 zmax = min 2 x⊥ x2⊥ and xa.

Consider two-body processes leading to final states consisting of partons. Phys.103 times larger than direct photon production. These events provide a testing ground for perturbative QCD. 579 (1984) [Addendum-ibid. (4. Eichten. which at LO describes two-body to two-body processes.4. 56.. 131 . The description of events with more than two jets requires higher-order calculations (which are beyond the scope of this course) that should account. E. Lane and C. at the parton level. The LO contribution to the cross section for direct production of photon pairs can then be estimated by scaling the dot-solid line in Fig. 1065 (1986)].80) for a factor of e2 /gs2 and then dividing by a factor of 2 to account for identical particles in the final state. K. Hinchliffe.80%.4. Of course. This effectively supresses the photon background by about two orders of magnitude while the signal efficiency remains between 70% . Quigg.g. The invariant amplitude for such a process can be simply obtained by multiplying Eq.11 by a factor of about 0. The LO contribution to diphoton reactions is given by the tree level process q q¯ → γγ. in the so-called tracker isolation criteria one defines a cone (in k⊥ and rapidity) around the direction of the photon. Rev.036. D. I. and therefore isolation cuts can be imposed to separate the hard scattering γ + jet topology.2 Two-Jet Final States Hard scattering processes in high-energy hadron-hadron collisions are dominated by events with most of the central hadronic activity concentrated in two jets. 58.22 The physical processes underlying dijet production in pp and p¯ p collisions are the scattering of two partons ij. Mod. producing two final partons kl that fragment into hadronic jets. 4. This background is mainly caused by events where high k⊥ photons are produced in the decay of neutral mesons or else are radiated from the quark (such as bremsstrahlung photons in the NLO QCD subprocesses). For example. 4. for the radiation which can occur from the initial and final state partons. the hadronic activity around the background photons tends to be much more than around the direct photons. with equal and opposite transverse momenta k⊥ 22 For a comprehensive description of multijet phenomena see e. and demands an absence of other particle tracks within that cone.

W ) fj (xb .4. (3.and p⊥ .35) we obtain |M|2 .4.2.92) where p2 = sˆ = (k ′ + p′ )2 = 2k ′ . 2ˆ s (4. W ) = dW 2 (2π)6 2E1′ 2E2′ ijkl × δ 4 (p − k ′ − p′ ) δ(p2 − W 2 ) 1 |M|2 . p′ = 2E1′ E2′ − kk′ p′k + p2⊥ .27) and (3.93) Using Eqs. The distribution of invariant masses W 2 = (k ′ + p′ )2 is given by Z Z 3 ′ Z 3 ′ XZ dσ dp (2π)4 dk dxa dxb fi (xa .1. and ′ δ 4 (p − k⊥ − p′⊥ ) = δ(E − E1 − E2 ) δ(pk − kk′ − p′k ) δ(~k⊥ + p~⊥ ) . respectively. (4.

.

1X 2 2 dσ 2 dσ .

4 spins dΩ dtˆ . |M| = 64π sˆ = 16πˆ s = .

and p′k = p⊥ sinh y2 .95) the cosine of the scattering angle in the parton-parton center-of-mass.ij→kl (4.97) 2 The integration over d3 k ′ d3 p′ can be conveniently rewritten in terms of jet rapidities y1 and y2 . 4.98) where y ≡ 21 (y1 −y2 ).4.4. 132 (4.96) and sˆ (1 + cos θ) .99) . kk′ = p⊥ sinh y1 . as sˆ tˆ = − (1 − cos θ) 2 (4.94) where the differential cross sections (dσ/dtˆ|ij→kl) for partonic subprocesses yielding jet pair production (shown in Fig. E2′ = p⊥ cosh y2 .12) are summarized in Appendix F.4. Since E1′ = p⊥ cosh y1 . (4. and their common transverse momentum: uˆ = d3 p π = dp2⊥ dy .4.4. The invariants may be expressed in terms of cos θ = (1 − 4 p2⊥ /ˆ s)1/2 . 2E 2 (4.4. (4. a straightforward calculation leads to E1′ E2′ − kk′ p′k = p2⊥ cosh(y1 − y2 ) ≡ p2⊥ cosh 2y .

12: Leading order Feynman diagrams for jet pair production.qi qi qi qi qi qi qj qj q¯j q¯j q¯i q¯i qi g g q¯i g qi q¯i g g qi g q¯i q (¯ q) q (¯ q) g q (¯ q) g g g g q (¯ q) g g g qi qj qi qi qi g q¯i q¯j q¯i q¯i q¯i g g qj q (¯ q) q (¯ q) g g g q¯j g g g g qi g g g g g q¯i g g g g g Figure 4. 133 .

4. W ) fj (xb .4. using the identity of hyperbolic functions.4.Now. W ) . δ p⊥ − δ(ˆ s − W ) = δ(4p⊥ cosh y − W ) = 4 cosh2 y 4 cosh2 y (4. we define sˆ W 2 4p2 τ= = = ⊥ cosh2 y (4.102) Eq. (4.92) becomes dσ π = (2πW 2) 2 dW (2π)2 Z dy1 Z dy2 XZ ijkl dxa Z dxb fi (xa .100) s s s so that   W2 1 2 2 2 2 2 . 1 + cosh 2y = 2 cosh2 y.101) Using Z Z 2 2~ 2 2 2 ~ d k⊥ d p~⊥ δ(k⊥ + p~⊥ ) δ(p⊥ − W /4 cosh y) = π dp2⊥ × δ(p2⊥ − W 2 /4 cosh2 y) = π.4. (4.

dσ .

.

4 cosh2 y dtˆ . 1 ′ ′ ′ ′ δ(E − E1 − E2 ) δ(pk − kk − pk ) × .

105) 134 . E1 ± kk′ = p⊥ e±y1 = p⊥ e±(Y +y) .104) da db δ(a) δ(b) = dA dB ∂(A. Y = 12 (y1 + y2 ). Putting all this together. (4.4. such that δ(a)δ(b) = Nδ(A) δ(B). the product of delta functions in Eq.4. (4. n√ o sx where E ± pk = √sxab .4.103) becomes √ δ(E − E1 − E2 ) δ(pk − kk′ − p′k ) = 2δ( sxa − 2p⊥ eY cosh y) √ × δ( sxb − 2p⊥ e−Y cosh y) √ √ = 2δ( sxa − W eY ) δ( sxb − W e−Y ) . and E2 ± p′k = p⊥ e±y2 = p⊥ e±(Y −y) . B) 2  ′ ′ The new variables can be rewritten as A B = E ± pk − (E1 ± kk ) − (E2 ± pk ).ij→kl (4. b) N δ(A) δ(B) = = 1 . with normalization N given by Z Z N ∂(a. (4.4.103) We now define a = E − E1 − E2 and b = pk − kk′ − p′k to perform the change of variables A = a + b and B = a − b.

13: Dijet invariant mass distribution in p¯ p collisions. yielding .8 TeV.Figure 4. at s = 1. The measurement is compared to a LO QCD calculation. and hence integration over the fraction of momenta is straightforward. as measured √ by the CDF Collaboration.

X √ √ −Y 1 dσ .

.

W ) fj ( τ e . dy1 dy2 fi ( τ e . Y . W ) cosh2 y ijkl dtˆ .

107) and the region of integration becomes |y1 | = |y + Y | and |y2| = |y − Y |. implying − ln(1/ τ ) < Y < ln(1/ τ ). xb < 1.4. y) (4.106) The Jacobian is found to be 1 dσ = Wτ dW 2 Z dy1 dy2 = ∂(y1 .ij→kl (4. Note √ √ that xa . y2) dY dy = 2 dY dy . The cross section 135 . ∂(Y.4.

per interval of W for pp → dijet can be rewritten in the form X Z 0 dσ dY fi (xa . W ) fj (xb . W ) = Wτ dW −Ymax ijkl .

Z ymax +Y dσ .

.

1 × dy .

W ) fj (xb . dtˆ ij→kl cosh2 y −(ymax +Y ) Z Ymax dY fi (xa . W ) + 0 # .

Z ymax −Y 1 dσ .

.

4.108) dtˆ . × dy (4.

95). The stated cuts on jet rapidities are equivalent to |y + Y |. with rapidity cuts |y1 |.23 The measurement is compared to a LO QCD calculation obtained through numerical integration of Eq. The data sample. Abe et al.13 shows the dijet invariant mass distribution as measured by the CDF Collaboration. the √ cut cos θ < 2/3 translates into a cut on the transverse momentum.108). collected with the Collider Detector at Fermilab. circles are from F. [CDF Collaboration]. setting cuts on jet rapidities. |y1|. Phys. when combined with the p⊥ cut further constrains the rapidity space: |y| < 0. xb = τ e−Y and the Mandelstam invariants occurring in the cross section are given by tˆ = − 12 W 2 e−y / cosh y. Figure 4.4. and sˆ = W 2 . |y2 | < 2. |y − Y | < 2. Using (4. [CDF Collaboration].8 TeV. and on the scattering angle in the dijet center-of-mass frame.4. Phys. ymax }. cos θ < 2/3. The cross section calculated at the partonic level using CTEQ6D PDFs and renormalization scale µ = p⊥ is normalized to the low energy data (180 GeV < W < 321 GeV) dividing the result of the calculation by 0. |y2| < 2. p⊥ > ( 5/6) W = 0. Rev.100) provides the relation W = 2p⊥ cosh y. The kinematics of the scattering (4. The CDF Collaboration made a precise measurement of the inclusive di√ jet differential cross section in p¯ p collisions at s = 1. (4.66. 23 Squares are from F.81.ij→kl cosh2 y −(ymax −Y ) √ √ where xa = τ eY . D 48. corresponds to an integrated luminosity of 106 pb−1 . Rev.4.4. D 55. Abe et al. The measurement is based on data binned according to the dijet invariant mass. The data distributions are in good agreement with LO QCD predictions.37 W . 136 . (4. The Y integration range in Eq. 998 (1993). which.108) √ is then Ymax = min{ln(1/ τ ). 5263 (1997). uˆ = − 12 W 2 e+y / cosh y.

4.14: Left Panel: Ratio of dijet invariant mass cross sections for rapidities in the interval 0 < |y1 |.|y2|<1. The Z + jet invariant mass spectrum is also shown.5 and 0. In terms of rapidity variable for standard transverse momentum cuts.0) 137 (4. The experimental points (solid circles) reported by the DO Collaboration are compared to a LO QCD calculation indicated by a dot-dashed line. To analyze the details of the rapidity space it is useful to introduced a new parameter.5) . Right Panel: LO QCD differential cross section as a function of dijet (γ + jet) invariant mass. cosh y = (1 − cos2 θ)−1/2 .12 QCD parton-parton cross sections are dominated by t-channel exchanges that produce dijet angular distributions which peak at small center-of-mass scattering angles. for y < 1 √ (y < 2. dσ/dW |(0.4) and s = 14 TeV.4. dijets resulting from QCD processes will preferentially populate the large rapidity region while the “new resonant” processes generate events more uniformly distributed in the entire rapidity region.5 < |y1|.5<|y1|.|y2|<0. (For details of the pp → Z + jet calculation see Appendix G). and the crossbar shows the size of the statistical error. R= dσ/dW |(|y1|. The error bars show the statistical and systematic uncertainties added in cuadrature. excitations of (hidden) recurrences result in a more isotropic distribution. |y2| < 1.109) . In contrast. |y2| < 0.Figure 4. As shown in Fig.

(4. For comparison we also show the invariant mass distribution of the photon + jet final state. in a given dijet mass bin.69 ±0.7 ±3.the ratio of the number of events.6 pb−1 for jet transverse energy thresholds of 30. 54. To accommodate the minimal acceptance from the LHC experiments an additional kinematic cut on the different jet rapidities. |y2| < 1.37.25 √ In Fig. has been included in the calculation. |y2| ≤ 1. as obtained from numerical integration of X Z 0 dσ dY fi (xa . 4.4.4.4. Figure 4.027.353 ±0. 85.14 shows the ratio R.5 and both rapidities 0. for both rapidities |y1|. as obtained through numerical integration of (4.0. and 91 ±5. |y1 |.14 we show the dijet invariant mass distribution at s = 14 TeV. and 115 GeV. The experimental points reported by the DO Collaboration24 (with integrated luminosities L = 0.109).108). 50. W ) = Wτ dW −Y max ijk . W ) fj (xb . 4.5 < |y1 |. respectively) are in good agreement with LO QCD calculation obtained through numerical integration of Eq. |y2| < 0.

Z ymax +Y 1 dσ .

.

× dy .

dtˆ ij→γk cosh2 y −(ymax +Y ) Z Ymax dY fi (xa . W ) + 0 # . W ) fj (xb .

Z ymax −Y 1 dσ .

.

110) × dy dtˆ . . (4.4.

the cross section for the inclusive process pp → dijet is about 2 to 3 orders of magnitude larger than pp → γ + jet. Goldberg. 101. 24 B. 2457 (1999). 82.ij→γk cosh2 y −(ymax −Y ) with the corresponding cuts on photon and jet rapidities. H. Anchordoqui. Phys. Lett. L. A. Abbott et al. Taylor. Nawata. 241803 (2008). Stieberger and T. D. Phys. R. Lett. Rev. 25 138 . Rev. As we anticipated in the previous section. Lust. S. [D0 Collaboration]. S.

(5. Hudson. Hoppes and R. D.1. Lee and C.Chapter 5 Precision Electroweak Physics 5. By analogy to the emission of photons in nuclear-γ decay. Rev. C. R.1) explained the properties of some features of β-decay. 11. Ambler. 161 (1934). Telegdi. E. 104. 254 (1956). D. Rev. Rev. Garwin. 88. 105. 1413 (1957). 105.2 Amazingly. Phys.1 The amplitude (5. Fermi considered the neutrino-electron pair to be created and emitted in the nuclear transition of a neutron to a proton. P. I. 105. known as the Fermi constant. T. J. Phys. Weinrich. R. Phys. needs to be determined by experiment. Friedman and V. D. L. 1415 (1957). Rev. L. attempts to unravel the true form of the weak interaction lead to a whole series of ingenious β-decay experiments. but not others. N.1 Charged and Neutral Currents The oldest and best-known examples of weak processes are the β-decay of atomic nuclei and the more fundamental neutron decay. Fermi. Wu. M. S. Lederman and M. Z. L. 2 139 . Phys. Inspired by the current-current form of the electromagnetic interaction he proposed that the invariant amplitude for the β-decay process be given by M = GF (un γ µ up ) (ν e γµ ue ) . 1681 (1957).1. W. the only essential 1 E. n → p¯ ν e− .1) where the effective coupling GF . Hayward. Over the following 25 years or so. 1 (1934). reaching the climax with the discovery of parity violation in 1956. Yang. Phys. Nuovo Cim.

The factor of 4 arises because the currents are defined with the normalized projection operator 21 (1 − γ 5 ) rather than the √ old-fashioned (1 − γ 5 ). Sudarshan and R. Sakurai. E. Therefore. 698 (1955). + (ν R . 193 (1958). Zh. (5.3) Weak interaction amplitudes are of the form 4GF M = √ J µ Jµ„ . Phys. e ).4) Charge conservation requires that M is the product of a charge-raising and a charge-lowering current.3 Fermi had not forseen parity violation and had no reason to include a γ 5 γ µ contribution. the charge-lowering weak current (1. 109. Nuovo Cim. J µ„ = [uν γ µ 21 (1 − γ 5 )ue ]„ = [u„ν γ 0 γ µ 12 (1 − γ 5 )ue ]„ = u„e γ 0 γ 0 21 (1 − γ 5 )γ µ„ γ 0 uν = ue γ 0 21 (1 − γ 5 )γ 0 γ µ uν = ue γ µ 12 (1 − γ 5 )uν . J. outgoing) lepton pair configura+ − 4 tions: (ν R . P. (5.1. The 1/ 2 is pure convention (to keep the original definition of GF which did not include γ 5 ).1. Phys. outgoing νL is the same as incoming ν¯R and viceversa. 7.1. G. Rev. This implies that the projection operator of the right-handed antiparticle is (1 − γ 5 )/2.2).g. B. 4 Recall that the spinor component of a right-handed antiparticle corresponds to the spinor component of a left-handed particle with negative energy.5. 649 (1958). e+ R ). J. the charge-raising weak current J µ = uν γ µ 12 (1 − γ 5 )ue (5. 0). Marshak.1. Feynman and M. E. viz.change required in Fermi’s original proposal was the replacement of γ µ by γ µ (1 − γ 5 ). 3 S. Further. C. Fiz.2) couples an ingoing negative helicity electron eL to an outgoing negative helicity neutrino. R. 29. 1860 (1958). S.1. Zel’dovich. and (eL ν R . e. Teor. νL eR ). Gershtein and Y. Besides the configuration (e− L . the charge-raising weak current also couples the following (ingoing. 140 . Gell-Mann. Rev. a mixture of γ µ and γ 5 γ µ automatically violates parity conservation. Eksp. 2 (5. 109. (0.118) is the hermitian conjugate of (5..2) represents a right-handed antineutrino ν R incoming and the right-handed positron e+ R outgoing.. νL ).

1.5 However.16637(1) × 10−5 GeV−2 . D. and hence universal. ν L and νR . etc) are some of the highlights in the development of particle physics. but Γ(π + → µ+ νL ) = Γ(π − → µ− ν¯R ) CP invariance . K decay. For example. the (1 − γ 5 ) form leaves the weak interaction invariant under the combined CP operation. N. ν denotes a muon neutrino.6) and muon lifetime are found to be within a few percent. can be prepared in intense beams which are scattered off hadronic targets to probe the structure 5 T. Comparison of these results supports the assertion that the Fermi constant is the same for all leptons and nucleons. (5. Although the experiments exposing the violation of parity in weak interactions (polarized 60 Co decay.136 ± 0. GF = (1. neutrinos.5) GF = 1. We discuss CP invariance in the next section. is a clear violation of parity invariance. 340 (1957). Lee. Rev. The absence of the “mirror image” states. It means that nuclear β-decay and the decay of the muon (see Appendix H) have the same physical origin. R. Yang. 106. Γ(π + → µ+ νL ) 6= Γ(π + → µ+ νR ) = 0 + + − − Γ(π → µ νL ) 6= Γ(π → µ ν¯L ) = 0 P violation .003) × 10−5 GeV−2 . The reason for the small difference is important and is discussed in the next section.The cumulative evidence of many experiments is that indeed only ν R (and νL ) are involved in weak interactions. In fact. particularly muon neutrinos. π decay.1. Also. these days. In this example. charge conjugation. Phys. (5. parity violation and its V − A structure can now be demostrated experimentally much more directly. The values of GF obtained from the measurements of the neutron lifetime. since C transforms a νL state into a ν¯L state. C. C violation . is violated . 141 . Oehme and C.

which we described in Chapter 4.7) and (5.1. Jqµ = uu γ µ 12 (1 − γ 5 )ud . For example.10) Upon inserting the currents (5.8) on the weak current the hermitian conjugates give the charge-lowering weak currents.9) (5.1. The procedure to embed the constituent cross sections in the overall νN inclusive cross section is familiar from Chapter 4: Z 1 Z xs CC d2 σνN . Quarks interact electromagnetically just like leptons.1.7) Jeµ = uν γ µ 12 (1 − γ 5 )ue .1. we obtain the invariant amplitude for the charged current (CC) neutrino-quark scattering.11) σ= dx dQ2 dxdQ2 0 0 142 . it is simplest to consider isoscalar nucleon targets. we model the charge-raising quark current. Our inclination therefore is to construct the quark weak current just as we did for leptons. 2 J + g √ Jµ 2 1 m2W  g √ Jµ„ 2  (5.1.1.of the weak interaction. This is analogous to the study of the electromagnetic lepton-quark interaction by scattering high-energy electron beams off hadronic targets.1. (5. we clearly need to know the form of the quark weak currents. The short range of the weak interaction results from the exchange of a heavy gauge boson of mass mW : J W g  =  = 4GF √ Jµ J µ„ . To predict the neutrino-quark cross sections. apart from their fractional charge. in which the nuclei contain equal numbers of protons and neutrons.1. To confront pQCD predictions with experiment. (5.8) into (5. N = (p + n)/2.10). (5.

Q2 ) + t¯s (x. and hence (5.12) is the differential cross-section given in terms of the structure functions. F2ν = x(u+d+2s+2b+ u¯ + d+2¯ 3 and FLν = 0.where  2   CC m2W G2F d2 σνN ν 2 ν 2 ν 2 = Y+ F2 (x. the structure functions are given in terms of parton distributions as ¯ c +2t¯). respectively. xF ν = x(u+d+2s+2b− u¯ − d−2¯ ¯ c −2t¯). we obtain  2   CC m2W G2F s d2 σνCC ¯N 2 = xq ν¯ (x. y = Q2 /sx. ν¯ xF3 = x(u + d + 2c + 2t − u¯ − d¯ − 2¯ s − 2¯b). Q ) + Y− xF3 (x. Q2 ) = u¯s (x.1. dy 143 (5. dxdQ2 4πx Q2 + m2W (5. Q2 ) + (1 − y)2xq CC ν (x.1. Q2 ) + dv (x. and s = 2Eν mN . Q ) − y FL (x. 2 2 dxdy π Q + mW (5.15) 2 the subscripts v and s label valence and sea contributions.1. Q2 ) + bs (x. dy dσνCC ¯N = c(1 − y)2 . (5.17) . Q2 ) + ds (x. Q ) . Q2 ) + d¯s (x. xF3ν¯ and FLν¯ .1.13) where uv (x.1. Q ) = CC dσνN = c.1. Q ) = 0. except for the replacement of F2ν .1. Q2 ) . t. 2 qCC ν (x.12) can be written in an “old hat” form  2 CC  CC  m2W G2F s d2 σνN 2 = xqν (x.16) CC 2 If there were just three valence quarks in a nucleon. That is. Q2 ) . xF3ν and FLν by F ν¯2 . Q ) . At leading order F2ν¯ = x(u + d + 2c + 2t + u¯ + d¯ + 2¯ s + 2¯b). and b denote the distributions for various quark flavors in a proton. (5. c. Q2 ) us (x. Going through the same steps. q¯ (x. d. s. Q2 ) + (1 − y)2xqνCC ¯ (x. Q2 ) + c¯s (x. with Y+ = 1 + (1 − y)2.14) qνCC (x. Y− = 1 − (1 − y)2 . At LO in pQCD. 2 2 dxdy π Q + mW (5. the neutrino-nucleon and antineutrino-nucleon scattering data would exhibit the dramatic V − A properties of the weak interaction. Q ) . Q2 ) + 2 2 + ss (x. The calculation of νN scattering proceeds along the lines of that for νN scattering. and u.

because of the (1 − y)2 behavior of the ν¯q cross section. Quigg. Gandhi. Sarkar. the NLO predictions for νN and νN CC inclusive cross sections are compared to those from a LO calculation using (5. JHEP 0801. Reno and I.13) and CTEQ4 PDFs. Earth may experience interactions with s ∼ however.1. Sarcevic.6 The calculation accounts in a systematic way for PDF uncertainties deriving from both model uncertainties and from experimental uncertainties of the input data set. A. Cooper-Sarkar and S. Phys. D. In Fig. the relation between the structure functions and the quark momentum distributions involve further QCD calculable coefficient functions. It is 144 . M. At low energies. QCD predictions for the structure functions are obtained by solving the DGLAP evolution equations at NLO in the MS scheme with the renormalization and factorization scales both chosen to be Q2 . M.1. Rev. Therefore. 093009 (1998). Rev. D 58. Recall that these equations yield the PDFs at all values of Q2 provided these distributions have been input as functions of x at some input scale Q20 . to obtain the structure functions. are difficult to test since the flux of cosmic neutrinos is virtually unknown. Rates for new physics processes. C. Sarkar. 043008 (2006). and contributions from FL can no longer be neglected. Hooper and S. D 74. where the contribution of the valence quarks predominates. 8 Ultrahigh energy cosmic neutrinos are unique probes of new physics as their interactions are uncluttered by the strong and electromagnetic forces and upon arrival at the √ > 200 TeV.13). and for the integrated cross sections 1 σνCC ¯N = . A. Phys. CC σνN 3 (5. H. the valence contribution is negligible and the νN and ν¯N cross sections become equal. the ν¯ cross sections are about a factor of 3 smaller than the corresponding νN cross sections. 7 R. Predictions for high energy νN CC inclusive cross sections have been calculated within the conventional DGLAP formalism of NLO QCD using the ZEUS-S global fit PDF analysis (updated to include all the HERA-I data). 5.1. A. Anchordoqui.7 The NLO results show a less steep rise of σ at high energies. Above Eν ≈ 106 GeV.8 6 L. 075 (2008).18) At NLO. reflecting the fact that more recent HERA data display a less dramatic rise at low-x than early data which was used to calculate the CTEQ4 PDFs.where c can be found from (5.1. Cooper-Sarkar. The resulting PDFs are then convoluted with coefficient functions.

JµNC (ν) = 1 2  uν γ µ 12 (1 − γ 5 )uν .94). are evidence of a weak neutral current.9 These events.g. 9 F. (5.21) where Tf3 and Qf are. D. A. compared with LO calculation. J. L.1). most readily interpretable as νµ (ν)N → νµ (ν)+ hadrons. Rev. see e. The neutral current interaction is described possible to mitigate this by using multiple observables which allow one to decouple effects of the flux and cross section.19) (5.. The discovery of neutrino-induced muonless events in 1973 heralded a new era in particle physics. Goldberg and A. respectively. they have right-handed components.1. Their values are cfV = Tf3 − 2 sin2 θw Qf cfA = Tf3 . Phys. 124027 (2002). In general.4. unlike the charged current Jµ . Shapere.1. (5. Phys. we see that the vector and axial-vector couplings.1.20) If we compare (5.  JµNC (q) = uq γ µ 21 (cqV 1 − cqA γ 5 )uq . H. L. D 65. the third component of the weak isospin and the charge of the fermion f (given in Table 2. Lett. are not pure V − A currents (cV 6= cA ). Anchordoqui. B 46. [Gargamelle Neutrino Collaboration].1. 145 . the JµNC .20) with (2.1: The NLO inclusive νN (left) and ν¯N (right) cross section along with the ±1σ uncertainties (shaded band). J. Hasert et al. Feng.NLO LO NLO LO Figure 5. cV and cA are determined in the standard model (given the value of sin2 θw ). 138 (1973).

For the moment we will leave ciV .24) while combining (5.23). Identification of (5.1. if the model is successful.1.4.26) In other words.1. 2 J NC + g JµNC cos θw 1 m2Z  g J NCµ„ cos θw  (5.1.1.25) from the last two equations and (2.1.1.10) yields GF g2 √ = . or by the ρ-parameter as can be seen by comparing (5.10) with (5. ciA and ρ as free parameters to be determined by experiment.1.22) and (5.1.1.1.1. 8m2W 2 (5.88) ρ= m2W = 1.22) with (5.9) and (5. = 8m2Z cos2 θw 2 (5. JNC Z g= cos w  =  = 4GF √ 2ρJµNC J NCµ„ .9) with (5.1.23) gives GF g2 √ ρ . For further discussion it is useful to remember that neutral currents have a coupling ρGF and that ρ represents the relative strength of neutral and charged weak 146 . all neutral current phenomena will be described by a common parameter. respectively.23) The relative strength of the neutral and charged currents is parametrized by the weak angle cos θw . m2Z cos2 θw (5.by a coupling g/ cos θw .22) (5.

(5.1. Q ) + ds (x. Q )   uv (x.29) and C 2 qN ν (x.1. Q2 )] [(cdV )2 + (cdA )2 ] + 2[cs (x.28) where the quark densities are given by    uv (x.and chargedcurrent amplitudes at low energy. Q2 ) + (1 − y)2 xq NC ν (x. for neutrino-quark scattering:  + Z t t + .. Q ) = (cV + cdA )2 + (cuV + cuA )2 2   2  us (x.30) ... The calculation of inclusive cross sections νN → νX proceeds exactly as that for the charged current processes. e. Q2 ) + dv (x. Q ) + bs (x.1. At LO in pQC we find NC d2 σνN G2 M Eν = F dx dy 2π  m2Z Q2 + m2Z 2   2 xqνNC (x. (5. Q2 )] [(cuV )2 + (cuA )2 ] . Q2 )  d NC 2 qν (x. Q ) + ds (x. 147 (5. Q2 )] [(cdV )2 + (cdA )2 ]  + 2[cs (x. Q2 ) + dv (x. Q2 )  d (cV − cdA )2 + (cuV − cuA )2 = 2   2  us (x. Q2 ) + ts (x. q ρ= (5.. Q2 )] [(cuV )2 + (cuA )2 ] . Q2 )  d 2 (cV ) + (cdA )2 + (cuV )2 + (cuA )2 + 2 2 2 + 2[ss (x.27)   + W q t b + .g. Q2 )  d 2 (cV ) + (cdA )2 + (cuV )2 + (cuA )2 + 2 2 2 + 2[ss (x. Q ) + bs (x. Q ) . Q2 ) + ts (x.1.currents. q0 ∆ρ measures the quantum corrections to the ratio of the neutral.

13) and (5. However. 90. (5.0007 . 5. 10 G. (5. Zeller et al.A quantitative comparison of the strength of NC to CC weak processes has been obtained by the NuTeV Collaboration.33) . (5.1. by scattering neutrinos off an iron target.2 Quark Flavor Mixing So far.10 The experimental values are Rνexp ≡ Rνexp ≡ ¯ σνNC µ N →νµ X σνCC µ N →µX σνNC ¯µ N →¯ νµ X σνCC ¯µ N →µX = 0.4050 ± 0.1. which only allows weak transitions between u ↔ d and c ↔ s. d µ− e− All these charged currents couple with universal coupling GF . This contradicts the above scheme. The K + is made of u and s¯ quarks. 88.30). Rev. we have seen that leptons and quarks participate in weak interactions through charged V − A currents constructed from the following pairs of (lefthanded) fermion states: ! ! ! u νµ νe .34) s formed from the heavier quark states. Phys. and . There must thus be a weak current which couples a u to an s¯ quark. It is natural to attempt to extend this universality to embrace the doublet ! c (5. 091802 (2002) [Erratum-ibid. the decay K + → µ+ νµ occurs. 239902 (2003)].1. P.2. using CTEQ4 PDFs. [NuTeV Collaboration].1.32) whereas for Eν > 107 GeV.2. For example. is Rν = Rν¯ ≃ 0. the prediction from (5.31) = 0.3916 ± 0.4.0016 . 148 . Lett. we already know that this cannot be quite correct.

11 In 1963. 531 (1963).Instead of introducing new couplings to accommodate observations like K → µ+ νµ . the data show that ∆S = 1 transitions are suppressed by a factor of about 20 as compared to the ∆S = 0 transitions. What we have done is to change our mind about the CC (5. For example Γ(K + → µ+ νµ ) ∼ sin2 θc .. Cabibbo first introduced the doublet u. s′ d′ + where d′ = d cos θc + s sin θc s′ = −d sin θc + s cos θc .. Rev. let’s try to keep universality but modify the quark doublets.1. the quark mixing angle. (5..2.0019. 10.36) This introduces an arbitrary parameter θc .2.35) . Γ(π + → π 0 e+ νe ) After allowing for the kinematic factors arising from the different particle masses.2255 ± 0. Cabibbo. (5. Phys. We now have Cabibbo favored transitions (proportional to cos θc ) u W+ c cos θc W+ d s and “Cabibbo suppressed” transitions 11 cos θc N. 149 . known as the Cabibbo angle. We assume that the charged current couples “rotated” quark states ! ! c u . This corresponds to sin θc = 0. Lett. . Indeed the mixing of the d and s quark can be determined by comparing ∆S = 1 and ∆S = 0 decays. d′ to account for the weak decays of strange particles.7). + + Γ(π → µ νµ ) Γ(K + → π 0 e+ νe ) ∼ sin2 θc .

which involves no mixing. (5. and c quarks.2. All this has implications for our previous calculations.1. we must eF = GF cos θc . their coupling to the third family.2. and similar diagrams for the charge lowering transitions.2. d. 2 b  150 (5. is unchanged.1.2.1. is very small.36)]. For example.1.38) The unitary matrix U performs the rotation (5.37) with γ µ (1 − γ 5 ) J µ = (¯ u c¯) U 2 d s ! . whereas the purely leptonic µ-decay replace GF in (5.2.39) gives a zeroth -order approximation to the weak interactions of the u.2. to embrace the additional doublet of quarks  d γ µ (1 − γ 5 )   J µ = (¯ u c¯ t¯) U  s .38). (5.2. We can summarize this by writing down the explicit form of the matrix element describing the CC weak interactions of the quarks.6) supports Cabibbo’s hypothesis.5) and (5. From (5. (5.5) by G rate.40) . The form (5. though non-zero.4) 4GF M = √ J µ Jµ„ 2 (5.2. The detailed comparison of these rates. (5. It is straightforward to extend the weak current. s.36) of the d and s quarks states: ! cos θc sin θc U= .u W+ c sin θc sin θc W+ s d [see (5.39) − sin θc cos θc Of course. there will be amplitudes describing semileptonic decays constructed from the product of a quark with a lepton current. J µ (quark) Jµ„ (lepton).

of Fig. 3 2 λ λ 1 |Utd | |Uts | |Utb | where λ = sin θc . This has fundamental implications concerning CP invariance.a. say the quark scattering process ab → cd.1.43) that is. L. Unlike the 2 × 2 matrix of (5. 49.2. The amplitude is „ µ M ∼ Jca Jµbd ∼ ∗ ∼ Uca Udb  „ u¯b γµ (1 − γ 5 )Ubd ud   u¯c γ µ (1 − γ 5 )ua u¯d γµ (1 − γ 5 )ub .The 3 × 3 matrix U contains three real parameters (Cabibbo-like mixing angles) and a phase factor eiδ . Phys. we first compare the amplitude for a week process. M′ = M„ . Rev. 51. Phys. 5. The approximation in (5. we 12 13 M. 151 . because of the phase δ. the amplitude M′ for the antiparticle process a ¯¯b → c¯d¯ (or cd → ab) is µ „ M′ ∼ (Jca ) Jµbd ∗ ∼ Uca Udb u¯a γ µ (1 − γ 5 )uc   u¯b γµ (1 − γ 5 )ud .1. By glancing back at (3. Wolfenstein. It is demanded by the hermiticity of the Hamiltonian.41) displays a suggestive but not well understood hierarchical structure. To investigate CP invariance.4).2. M describes either ab → cd or c¯d¯ → a ¯¯b (remembering the antiparticle prescription of Sec.2.2. 652 (1973). Prog. Theor. with that for an antiparti¯ We take ab → cd to be the charged current interaction cle reaction a ¯¯b → c¯d. Kobayashi and T. Lett. 1945 (1983).41) U =  |Ucd | |Ucs | |Ucb |  ∼  λ 1 λ2  .3) and (3. each element may be multiplied by a phase and a coefficient of O(1). (5.11).39).42) „ ∗ because Ubd = Udb . Maskawa.13 These are order of magnitude only. This should not be surprising. 1. which we discuss next.2.2.12 An easy-to-remember approximation to the observed magnitude of each element in the 3-family matrix is     1 λ λ3 |Uud | |Uus | |Uub |     (5. the CabibboKobayashi-Maskawa (CKM) matrix is complex. The original parametrization was due to Kobayashi and Maskawa. u¯c γ µ (1 − γ 5 )Uca ua (5. On the other hand.

we need u¯C and also. . then the theory is CP invariant.1.2. then is CP violated.2. (5.45) see Sec.4. How do we verify that the theory is CP invariant? We calculate from M(ab → cd) of (5. In the standard representation of gamma matrices we have „ u¯C = uC γ 0 = (Cγ 0 u∗ )„ γ 0 = uT γ 0 C „ γ 0 = −uT C „ γ 0 γ 0 = −uT C −1 . but have indicated that invariance under the combined CP operation may hold. . Clearly to form MCP .2. see that M is essentially the interaction Hamiltonian V for the process. 5. In Sec.2.44) where uC are charged conjugate spinors defined by uC = C u¯T . If it does not.47) . if MCP = M„ . i = a. 1. to know how γ µ (1 − γ 5 ) transforms under C. where M describes the i → f transition and M„ describes the f → i transition in the notation of Chapter 3.2. (5.42) ui → P (ui)C . describing the CP -transformed process.2.42) the amplitude MCP . d .46) γ µ = −(Cγ 0 )γ µ∗ (Cγ 0 )−1 = −Cγ 0 γ µ∗ γ 0 C −1 = −Cγ µT C −1 152 (5. If it does. and see whether or not the Hamiltonian remains hermitian. The total interaction Hamiltonian must contain M + M„ .2: The processes described by (a) the weak amplitude M(ab → cd) and (b) its hermitian conjugate. we have seen that weak interactions violate both P invariance and C invariance. . (5. that is.a c Uca a W+ b ∗ Udb c ∗ Uca a ¯ W− d b Udb W− ≡ ¯b d c¯ ∗ Uca Udb d¯ (b) (a) Figure 5. MCP is obtained by substituting the CP -transformed Dirac spinors in (5.

the first charged current of (5. (5. and the theory is CP invariant. As a result. we find MCP = M„ .42) becomes µ C (Jca ) = Uca (¯ uc )C γ µ (1 − γ 5 )(ua )C = −Uca uTc C −1 γ µ (1 − γ 5 )C u¯Ta = Uca uTc [γ µ (1 + γ 5 )]T u¯Ta = (−)Uca u¯a γ µ (1 + γ 5 )uc . antiparticles states are not allowed.49) The origin of the extra minus sign introduced in the last line is subtle but important. Thus µ (Jca )CP = (−)Uca u¯a γ µ „ (1 − γ 5 )uc .48) With the replacements (5.50) and hence    ∗ MCP ∼ Uca Udb u¯a γ µ (1 − γ 5 )uc u¯b γµ (1 − γ 5 )ud .2. in a single-particle theory. The parity operation P = γ 0 .2.44).2. (5. and the formalism is completely f ⇔ f¯ symmetric.2. 153 . The minus sign is related to the connection between spin and statistics.2.2.2. and so P −1 γ µ (1 + γ 5 )P = γ µ „ (1 − γ 5 ). However. (5. (5.43). in field theory it occurs because of the antisymmetric nature of the fermion fields.C −1 γ µ γ 5 C = −γ µT C −1 iγ 0 γ 1 γ 2 γ 3 C = −iγ µT (C −1 γ 0 C)(C −1 γ 1 C)(C −1 γ 2 C)(C −1 γ 3 C) T T T = −iγ µT γ 0 γ 1 γ 2 γ 3 T = −γ µT (iγ 3 γ 2 γ 1 γ 0 )T = −γ µT (iγ 0 γ 1 γ 2 γ 3 )T = −γ µT γ 5 T = −(γ 5 γ µ )T = (γ µ γ 5 )T . In field theory. the charge conjugation operator C changes a positive-energy particle into a positive-energy antiparticle.51) We can now compare MCP with M„ of (5. rather C changes a positive-energy particle state into a negative-energy particle state. we must add to our Feynmann rules the requirement that we insert by hand an extra minus sign for every negative-energy particle in the final state of the process. Provided the elements of the matrix U are real.

Sci. Christenson.52) J. in general. L. J. has been established in the decay Bd → Kπ with a significance in excess of 5σ.3. A. Vives.3 Scalars were already part of the Theory! One can illustrate this statement simply by calculating the cross section for top quark annihilation into Z’s.2. 51. Rev. Fitch and R. Gonzalez-Garcia. 154 . Phys. In fact. C. 4861 (1995). s (5. V. It now contains a complex phase factor eiδ .15 5. However. 13.14 These particles offer a unique “window” through which to look for small CP violating effects. we have MCP 6= M„ and the theory neccesarily violates CP invariance. Given the rapid progress and the better theoretical understanding of the standard model expectations relative to the K system. not mixing-assisted. precision data on neutral kaons have been accumulated over 40 years. Therefore. the hope is that at this point. 138 (1964). Stelzer and R. s ≫ mt . s) level. A. the matrix U becomes the 3 × 3 CKM matrix. H. The evidence for the indirect violation of CP -invariance was first revealed in 1964 in the mixing of neutral kaons. Ann. In the energy limit. (5. in a standard model without √ scalars. tt¯ → ZZ. 161 (2001). Turlay. Lett. 15 F. Masiero and O.At the four-quark (u. Cronin. Vazquez. straightforward Feynmanology yields  dσ   dΩ t t  crossed  α2 m2t +O + = diagram m4Z 14   1 . the results have fit snugly within the standard model.39). Nucl. Halzen. with the advent of the b and t quarks. M. Part. direct CP -violation. a tiny CP violation had been established many years before the introduction of the CKM matrix. without exception. W. Today. d. Phys. Rev. Rev. thus far. the measurements can. c. In particular. this is the case as the 2 × 2 matrix U. T. be accommodated by the standard model with three families. the glass is half full and that improved data will pierce the standard model’s resistant armor. Whenever the experimental precision in CP -violation measurements has increased. is indeed real. D 51.

3.3.3. 16 C.54).3. if scalars were not invented to solve the problem of mass.3.55) which.3. (5.3.3. Lett.3.3. m4Z (5. s (5.53) where s is the square of the tt¯ annihilation energy. (5.55) Here δJ are the phase shifts. Phys. We therefore have to conclude that the process violates S-wave unitarity. when combined with (5. Llewellyn Smith.We first notice there is no angular dependence. they would have to be introduced to salvage unitarity. B 46.56) and (5.16 The cancellation requires that the top-Higgs coupling Yt (endowing the top quark with mass) satisfies Yt2 ∼ m2t . yields σJ < 16π(2J + 1) 1 s (5.53) represents the special case J = 0. The process is purely S-wave. dσ/dΩ is independent of Ω. 233 (1973). The Higgs particle comes to the rescue.3. introducing the additional diagram: t Z H Yt  mt t ∝ Yt2 . which requires that σJ=0 ∼ 1 .58) a result indeed intrinsic to the Higgs origin of fermion masses.52). We remind the reader that the unitarity constraint (5.57) Z which cancels the ill-behaved J = 0 term (5.53) simply follows from the partial wave expansion of the cross section in ordinary quantum mechanics: 16π X (2J + 1) |fJ |2 . (5.54) σ= s J with fJ = exp(iδJ ) sin δJ . H. So. Obviously |fJ |2 < 1 from (5. 155 .

59).3. a requirement which follows from the recognition that the standard model’s perturbative predictions are correct. Tevatron data excluded the mass range (160 GeV.63) with Note that the velocity v is the variable which mixes electric and magnetic interactions.59) The lower limit can be deduced from unsuccessful searches. 114.3. This requires couplings to be small.3. Very recently. Hence our 1 TeV value quoted in (5.62) e = eM .4 GeV < mH ∼ (5.4 Electroweak Model @ Born Level Some 150 years ago Maxwell unified the electric and magnetic forces by postulating the identity of the electric and magnetic charges: ~ + eM ~v × B ~.60) yields the upper limit mH = 2λv 2 1/2 < √ 2 v ≃ 350 GeV . but we know that < 1 TeV . when v → 0 magnetic interactions are simply absent but. 156 .3. Unification of the two forces introduces a scale in the mixing variable v: the speed of light.17 The vacuum expectation value v2 = 1 1 = (246 GeV)2 4m2W = √ 2 g 2GF (5. (5.61) The inequality follows from λ < 1.4 GeV. the two interactions become similar in importance.4.4. an argument which cannot be taken too literally as it cannot distinguish λ < 1 from λ < 4π. for charges moving with significant velocity v. 170 GeV) at 95% CL. 5. F~ = eE (5. 17 The combination of LEP data yields a 95% CL lower mass of 114.We have not found the Higgs particle. for instance. (5.

The SLD experiment at the SLC at SLAC observed some 5 × 105 events. DELPHI.64) contains a parameter θw which is left to be determined by experiment.18 We have barely started down the road of high precision tests familiar from quantum electrodynamics. just like the electric and magnetic forces are in the interaction of high velocity charges. and OPAL at CERN observed ∼ 2 × 107 Z bosons. During the period 1989-1995 the four LEP experiments ALEPH. L3. 18 The first phase of the LEP/SLC program involved running at the Z pole. Not until the mid-ninties did true verification of the electroweak theory become possible with the first confrontation of its calculated radiative corrections with high statistics measurements performed at the LEP and SLC e+ e− colliders and at the p¯ p Fermilab Tevatron.22) and recall (5. What is the variable mixing electromagnetic and weak forces? At low energy the effects of weak forces between charged particles are swamped by their electromagnetic interaction.4. The energy scale introduced by their unification is the weak boson mass mW .64) generalizes (5.1. The first and only tangible confirmation of electroweak unified theory has been provided by verification that the ratio of the weak boson masses determined at proton-antiproton colliders yields a value of the weak angle which is in agreement with the value determined in the pioneering neutral current neutrino experiments.22). At a modern accelerator the weak and electromagnetic forces are equally obvious in the collisions of high energy particles. with center-of-mass energy gradually increasing from about 140 GeV to 209 GeV. We describe the first successful steps next.4.88).64) expressing the equality of electric and weak charge g in terms of the parameter θw introduced in (5. The sad reality is that electroweak unification (5. see (2. On a more mundane level. e+ e− → Z → ℓ+ ℓ− . Energy is the mixing variable of electromagnetic and weak forces. q q¯.4.4. In the electroweak theory (5.9) and (5.4.27)) as well as the ratio of the weak boson masses mW and mZ .1. LEP2 ran from 1995-2000. 2. and ν ν¯.88).4 via (2.63) to include the weak force.4. 157 . (5. this common value verifies the doublet nature of the scalar field introduced in Sec. The parameter represents the relative strength of charged and neutral currents (cf. (5.Unification of the electromagnetic and weak interaction follows this pattern with e = g sin θw .1.1.

For PETRA e+ e− beam energies ∼ 20 GeV we have k 2 ≈ s ≈ (40 GeV)2 and so predicts about a 15% effect.4.65) using GF ≈ 10−5 /m2N [see Appendix H] and e2 /4π = 1/137. To make a detailed prediction. As we discussed in Sec. 5.4. 3. at least by QED standards. Perhaps. Using Feynman 158 . collider and fixed-target experiments with lepton.3. In the electroweak sector familiar tests of the standard model probe the Lagrangian at Born level.3: Electromagnetic and weak contributions to e+ e− → µ+ µ− . with a small weak contribution. Therefore. e+ e− annihilations can occur through electromagnetic (γ) or weak neutral current (Z) interactions. In what follows. which is readily observable. The size of this effect is found to be GF 10−4 k 2 |MEM MNC | ≈ ≈ . photon. we assume that the neutral current process is mediated by a Z boson with couplings given by (5. the measurement of the reaction e+ e− → µ+ µ− at PETRA energies provides tests of the validity of QED at small distances.1 Interference in e+ e− annihilation When contemplating the vast amount of evidence for the standard model. we show that such a measurement also provides a unique test of the asymmetry arising (in the angular distribution of muon pairs) from the interference of the electromagnetic amplitude MEM ∼ e2 /k 2 . high-energy e+ e− colliding beam machines are an ideal testing ground for the interference effect of the electromagnetic and the neutral weak amplitude.µ+ e+ γ e− µ+ e+ Z µ− e− µ− Figure 5. it is easy to overlook the fact that verification of the theory at the quantum level is in its infancy. the oldest of these tests has been the measurement of electroweak interference in e+ e− collisions. and hadron beams. |MEM |2 e2 /k 2 m2N (5.20). covering strong and electroweak interactions.1.

With electronmuon universality. the amplitudes Mγ and MZ corresponding to the diagrams of Fig. . and where cR ≡ cV − cA . It is easier to calculate |Mγ + MZ |2 in this form. 5. which enable MZ to be expressed explicitly in terms of right. (5.4.4. s ≃ k 2 . the superscripts on cV.rules.4.1. With definite electron and muon helicities.A are superfluous here.70) The (1 ± γ 5 ) are projection operators.4. but we keep them so as one is able to translate the results directly to e+ e− → q q¯. For example.67) becomes √ 2GF m2Z µ [cR (¯ µR γ ν µR ) + cµL (¯ µL γ ν µL )] [ceR (¯ eR γν eR ) + ceL (¯ eL γν eL )] . MZ = − s − m2Z (5. 3.69) That is we have chosen to write cV 1 − cA γ 5 = (cV − cA ) 12 (1 + γ 5 ) + (cV + cA ) 21 (1 − γ 5 ) .26) with ρ = 1.68) using (5. (5. so the Dirac equation for the incident positron reads ( 12 kσ )eγ σ = 0 and the numerator of the propagator simplifies to gµσ . cL ≡ cV + cA .24) and (5.3. (5.4.67) where k is the four-momentum of the virtual γ (or Z). We ignore the lepton masses.1.and left-handed spinors. k MZ    ν µ  gνσ − kν kσ /m2Z g2 µ 5 = − µγ (cV 1 − cA γ )µ 4 cos2 θw k 2 − m2Z  σ e  × eγ (cV 1 − ceA γ 5 )e .3 are e2 (5. we can apply the results of the QED calculation of e+ e− → µ+ µ− given in Sec. Thus.66) Mγ = − 2 (µγ ν µ)(eγν e) .4. (5.

α2 dσ .

.

4. = (5. (1 + cos θ)2 [1 + rcµL ceL ]2 .71) .

dΩ e− e+ →µ− µ+ 4s L R L R .

dσ .

.

. α2 = (1 + cos θ)2 [1 + rcµR ceL ]2 .

dΩ e− e+ →µ− µ+ 4s L R R L 159 (5.4.72) .

|r| ≪ 1)   A1 2 3c2A GF s 2 AF B = . ≃ ℜe(r)cA ≃ − √ (8A0 /3) 3 e2 2 (5. of (5.65).66). We find. Let us calculate the size of the integrated asymmetry defined by Z 1 Z 0 dσ dσ F −B with F = dΩ .77) AF B ≡ F +B 0 dΩ −1 dΩ Integrating (5.72) hold for the other two nonvanishing helicity configurations. (5.e. (5.74) = dΩ 4s where.4.[see (3.74). To calculate the unpolarized e+ e− → µ+ µ− cross section.4. B= dΩ .4.4. that is. (5. Here.4. which is important for s ∼ m2Z [see Appendix A]. an asymmetry which grows quadratically with the 160 .4. Expressions similar to (5.76) The lowest-order QED result (A0 = 1.75) (5.68) and (5.4.78) This is in agreement with the expectations of the order of magnitude estimate.4..  α2  dσ A0 (1 + cos2 θ) + A1 cos θ . (5.4. A1 = 0) gives a symmetric regular distribution. We now see that the weak interaction introduces a forwardbackward asymmetry (A1 6= 0). r is the ratio of the coefficients in front of the brackets in (5.4.73) r= s − m2Z + imZ ΓZ e2 where we have included the finite resonance width ΓZ . (assuming electron-muon universality cµi = cei ≡ ci ) A0 ≡ 1 + 21 ℜe(r)(cL + cR )2 + 41 |r|2(c2L + c2R )2 = 1 + 2ℜe(r)c2V + |r|2(c2V + c2A )2 .3.4. GF s/e2 . √ s 2GF m2Z .4. R helicity combinations. we obtain for s ≪ m2Z (i.51)]. we average over the four allowed L. A1 ≡ ℜe(r)(cL − cR )2 + 12 |r|2(c2L − c2R )2 = 4ℜe(r)c2A + 8|r|2c2V c2A .71) and (5.

4: The e+ e− → µ+ µ− angular distribution for all CELLO data √ h si = 43 GeV. The cos θ distribution does not follow the 1 + cos2 θ QED prediction. 161 .Figure 5.

energy of the colliding e+ e− beams (for s ≪ m2Z ). Q2 ) + xq(x.4.74) with the experimental measurements of the high-energy e+ e− → µ+ µ− angular distribution. LO analytic expressions for the strength of NC to CC weak processes can be easily obtained from (5. We may use the standard model couplings (cA = 21 . Q2 ) ¯ = .80) (5. r Rν¯ = where r≡ σν¯ℓ N →ℓX 3x¯ q (x. offer an accurate determination of sin2 θw . these are given by an isoscalar target (q = u+d 2 (3gL2 + gR2 )xq(x. 3. for ).1.1. Q2 ) 1 = gL2 + gR2 . 9 gL2 ≡ (gLu )2 + (gLd )2 = (5. however. gRq ≡ 12 (cqV − cqA ).13) and (5.83) and gLq ≡ 21 (cqV + cqA ).79) and (3gR2 + gL2 )xq(x.2 The NuTeV anomaly Neutral current processes in deep inelastic neutrino-nucleon scattering provide a direct measurement of the electroweak mixing angle. Q2 ) + (3gL2 + gR2 )x¯ q (x. Q2 ) xq(x. Including only first generation quarks. Q2 ) + 3x¯ q (x.8. cV = − 12 + 2 sin2 θw ≃ 0) to compare (5.81) (5.4. Q2 ) + (3gR2 + gL2 )x¯ q (x. σνℓ N →ℓX 3xq(x. Since cV ≃ 0. Q2 ) + x¯ q (x. Q2 ) + x¯ q (x. 5. Q2 ) 3xq(x.4.4.82) (5. Q2 ) = gL2 + rgR2 Rν = (5. these data do not. 5.4. 162 (5.4.84) .30). we see in this case the larger statistics clearly reveal the data are inconsistent with QED predictions.4. see Fig.4. 2 9 5 gR2 ≡ (gRu )2 + (gRd )2 = sin4 θw . Compared to the results shown in Fig.4. Q2 ) 5 1 − sin2 θw + sin4 θw .

(5.19 exp Actually.4. Wolfenstein. primarily because the neutral current scattering of ν and ν¯ yield identical observed final states which can only be distinguished through a priori knowledge of the initial state neutrino. the contamination of the muon neutrino beam by electron neutrinos.80). of every 100 experiments you expect about one 3σ effect. the NuTeV data can be viewed as a measurement of the ratios between the CC and the NC squared neutrino effective couplings. and the efficiency of NC/CC discrimination. Furthermore. On the theoretical side.85) which is seen to be independent of q and q¯. The electroweak parameter sin2 θw extracted from a single parameter fit to the NuTeV data is about 3σ at variance with the overall fit of the standard model to precision observables.4. RPW is more difficult to measure than the ratio of the neutral current to charged current cross sections.4. Rev. the measurement of RPW requires separated neutrino antineutrino beams.20 Once all these effects are taken into account. 20 163 . (5. D 7. as well as QCD and electroweak corrections. 91 (1973).The difference of the effective couplings gL2 − gR2 is subject to smaller theoretical and systematic uncertainties than the individual couplings. the NuTeV measurement is 19 E.79).” A 3σ effect is not neccesarily cause for excitement. such as those related to the spectrum of the neutrino beam. On the experimental side. RPW measured at NuTeV differ from the expressions given in (5. and therefore of the information on the partonic structure of the nucleon.4. this is because of contributions from second–generation quarks. Therefore. A. Indeed. Rνexp ¯ . Phys. and (5. this is because total cross sections can only be determined up to experimental cuts and uncertainties. under the assumptions of equal momentum carried by the u and d valence quarks in the target and of equal momentum carried by the heavy quark and antiquark seas.85). a fact that is known as “the NuTeV anomaly. As a matter of fact. we obtain Rν − rRν¯ 1−r σνN →νX − σν¯N →¯ν X = σνN →ℓX − σν¯N →ℓX ¯ 2 2 = gL − gR RPW ≡ = 1 2 − sin2 θw . the observables Rνexp . Paschos and L.

. Nucl. Sirlin. and ∆κ. 905 (1978)]. 84 (1983). ∆r. Marciano. 50. taking into account all these considerations and their uncertainties. For a wide class of low-energy and Z-boson observables. 192001 (2007). D. Hollik. G. S. Phys. 89 (1977). Phys. Phys.. Sirlin. Sirlin. 50. Phys. In Z-boson physics. A. Degrassi. cfV /cfA = 1 − 4|Qf | sin2 θ¯w . the dominant effects arise from light charged fermions. B 351. F.g. 135 (1975).22 These parameters bear the following physical meanings: (i) ∆α determines the running fine structure constant at the Z boson scale α(mZ )/α = (1 − ∆α)−1 . Rev. Phys. Phys. J. D 22. e.5 Radiative Corrections As a rule. Veltman. W. 573 (1978) [Erratum-ibid. Phys. GF m4t . J. Veltman. etc. Phys. i. Phys. S. G. 38. Lett. electroweak radiative corrections.. is yet to see the light of day. L. (ii) ∆ρ measures the quantum corrections to the ratio of the neutral. (iv) ∆κ controls the effective weak mixing angle. sin2 θ¯w = sin2 θw (1 + ∆κ). NLO QCD effects. Rev. (iii) ∆r embodies the non-photonic corrections to the muon lifetime. which generates power corrections of the orders GF m2t . the dominant effects originate entirely in the gauge boson propagators (oblique corrections) and may be parametrized conveniently in terms of four electroweak parameters: ∆α. 123 (1989). Nucl. B 95. A full re-analysis of the data.and charged-current amplitudes at low energy. and nuclear shadowing. S. J. 22 164 . that occurs in the ratio of the Zf f¯ vector and axial-vector couplings. A. Rev. 971 (1980). M. Sirlin.21 Other possible systematic effects that could contribute to bridge the gap are large isospin violation in the nucleon sea. which induced large logarithms of the form αn lnm [m2Z /m2f ] (with m ≤ n) in the fine structure constant. A. the size of radiative corrections to a given process is determined by the discrepancy between the various mass and energy scales involved.e. 5. 319 (1990). the ∼ 3σ result is reduced to ∼ 2σ if one incorporates the effects of the difference between the strange and antistrange quark momentum distributions. A. 165 (1990). Sarantakos. Mod. B 232. Fanchiotti and A. Ross and M. Rev. Nucl. D 41. B 123. 49 (1991). Fanchiotti and A. Sirlin and W. G. Nucl. Fortsch. and from the top quark. Lett.fraught with hadronic uncertainties. The ensuing discussion contains an aper¸cu of the theory of electroweak 21 D. Mason et al. 99. ∆ρ. αs GF m2t . Phys. B 217.

5. mZ 3 1/2  4 ≃ 2(1 − 4s2 ) .91) represent an incomplete list of experiments capable of measuring sin2 θw .and right-handed electrons. For 165 . (5. α. and (v) the various asymmetries (5. predicting the top quark mass.87)– (5. m2Z ≡ z . σ(¯ νµ e) 1 − 4s2 + 16s4 (5. m2W ≡ w . Introducing the notation sin2 θw = s2 = 1 − c2 .5.5.5.5.5. 2 GF w (5.radiative corrections and its role in testing the standard model.87)–(5.91) measured at Z-factories (see Appendix I).87) w = 1 − s2 . (ii) the measurement of the weak boson masses (5. (iii) the combination of mW .5. constraining the Higgs boson mass.5.5. which is a function of sin2 θw only.87) of νµ scattering on left.88). After inclusion of the O(α) corrections. Validity of the standard model requires that each measurement yields the same value of s2 : (i) the ratio (5.5. z (5.87)–(5. and color factor CF = 3 (1) for quarks (leptons).5. and searching for deviations that may signal the presence of new physics. ALR ≃ Aτ ≃ AFB 3 (5.5. Implementing such a program can be first formulated from the point of view of the experimentalist.5.90) (5.5.88) πα 1 √ = s2 . (iv) the partial widths (5.91) in different ways.5.90) of the Z into a fermion pair with vector and axial coupling cfV and cfA .89).91) is not straightforward.91) Equations (5. and GF as determined by the muon lifetime (5. the sin2 θw values obtained from the different methods will no longer be the same because radiative corrections modify (5.5.86) electroweak theory predicts at the Born level that: 3 − 12s2 + 16s4 σ(νµ e) = . The study of the quantum corrections to the measurements (5.5.89)   ¯ Γ(Z → f f) α = CF (cfV )2 + (cfA )2 .

. The experimentalist has to make a choice and define the Weinberg angle to O(α) by one of the observables (5. Z masses f Z W + +. Subsequently. all other experiments should be reformulated in terms of the preferred “sin2 θ.5.5.87)–(5. m2Z 166 (5. There is no real mystery here. contribute to O(α) shifts in the W. however.5.27).88). After inclusion of O(α) contributions in Eqs.92) W e W e e modifies the t-channel Z propagator measured by (5.e.5. i.93) f W H + +.1.95) .5. + f W (5.5.23122(15) .example. + f W (5.. see also (5.94) which yield an improved sin θw value via (5.87).5.87)–(5. . 2 sin2 θw ≡ 1 − m2W = 0. they represent different definitions of sin2 θw . .5.5. (5. It does not. the diagram  Z (5.” What this choice should be is no longer a matter of debate and we will define sin2 θw in terms of the physical masses of the weak bosons.91).5.91).

g. in a straightforward way to the electroweak model. We pay a price: the Thomson charge is no longer predicted and the charge is renormalized to its measured value at q 2 = 0.5. mf .95). now includes α. 3. Yf .5. Renormalization techniques take care of UV divergences appearing in gauge theories at the quantum level.5. (5.5. .89). which generalizes the renormalization techniques.5. which has been previously introduced. mZ . mH .5. The choice (5. . the parameters of the minimal Higgs potential.90) and (5.5 we illustrated how the divergence in the photon vacuum polarization is absorbed into the Thomson charge.96) where mf represents the lepton and quark masses me . mW .95) is particularly useful in that one can estimate the radiative corrections in terms of the renormalization group.5. For some this procedure may seem unfamiliar. In a more technical sense the choice (5.91). (5. The screening of the charge α(q 2 ) can still be predicted and confronted with experiment. . (5. mt .97) which represent the bare electroweak couplings. .5. µ. The same procedure can be repeated for the other measurements of θw . (5. The O(α) corrections can be qualitatively understood in terms of the loop corrections to the vector-boson propagators (5. introduced for electrodynamics.5. e.5.A most straightforward test of the theory is now obtained by fixing (5.87) written in terms of (5. α and me . mZ via (5.93) and (5. All UV divergences in QED can be absorbed in two parameters. It is eminently reasonable to copy this scheme for calculations in electroweak theory. λ. In Sec.95) in terms of the measured weak boson masses and verifying that its value coincides with the value of sin2 θw obtained from an analysis of ν deep-inelastic scattering data using the O(α) prediction for (5. g ′. Not all predictive power is lost.88). its value is automatically determined by mW . The weak mixing angle sin2 θw does not appear in the list of parameters.5.94). Traditionally the standard model Lagrangian is determined in terms of g. and the “Yukawa” couplings of the Higgs particle to 167 .5.. to be fixed by experiment.5.95) is closely related to the use of the on-mass-shell (OMS) renormalization scheme. The list of parameters.

5. The origin of the relation (5. .89) is calculated to O(α) in terms of the weak angle θw defined by (5.fermions.5. 2 (5.97) z .5.89) is the muon’s lifetime which. 168 (5.88).101) g 2 = e2 g ′2 = e2 λ = e2 and 2 As an example we will show how the relation (5.5. There is.99) zm2H . z−w (5.96) and (5. to leading order. There is no mystery here.98) z . in fact.5. electromagnetic radiative corrections must be included to obtain the result to O(α).104) .5.100) zm2f .103) where photonic corrections = +.5. 8w(z − w) (5.5.5.5.5. w (5. GF Γ(1) µ = √ [1 + photonic corrections] .102) W  In Fermi theory.5. Symbolically. In principle any choice will do. . Yf = e 2w(z − w) (5. a direct translation between sets (5. is given by the diagram  Γ(0) µ = e e (5.

5. . 2 ws2 169 (5.107) and Z box = W +.105) + box where f t + propagator = b f H +.  (5. on the other hand.5. + (5.105) we obtain πα 1 GF = √ (1 + ∆r) . . . .In electroweak theory. (5. . e2 h (1) 1 + photonic corrections Γµ = 2 2 8s c z + propagator + vertex i (5.106)  vertex = Z0 +.5.5.109) .5.5. .5.103) and (5.108) Equating (5.

110) recognizes the fact that in the OMS scheme.5. Stong. Roy and M. Phys. Lett. B 353. Part.112) ∆ρ = W b Other contributions are small in the OMS scheme and are grouped in the “remainder” ∆rem . C 58. 233 (1994). M. B. Before discussing the status of measurements of ∆r.23 To the extent that ∆rem is small. B 277. Halzen and D. Notation (5.110) We note that the purely photonic corrections drop out.87) and (2. Vazquez. F.5.111) f f as well as the third generation. Nucl. heavy quark diagram t (5.5. Morris. vacuum polarization loops dominate this quantity.5. Phys. 10 (1991). 170 . 107 (1990). Halzen.with ∆r = ∆α − c2 ∆ρ + ∆rem .5. B 322. A. World 2.5. Kniehl. A. 567 (1991). Phys. Gonzalez-Garcia. 119 (1993).5. s2 (5. f ∆α = X (5. one 23 F. The full order α calculation of ∆r will not be presented here. Phys. P.89). (5. Stong. Lett. Kniehl and M. F. Z. Halzen.88). L. Lett. To leading order ∆r = 0 and. F. C. Phys. We have attempted to describe the full formalism in a relatively accessible way elsewhere. we make several comments. the electroweak radiative corrections are gathered in ∆r.4. A. As mentioned above. 503 (1992). using (3. L. B 237. Halzen and B. We specifically isolated the fermions which are responsible for the running of α from the muon to the Z mass. A.109) reduces to the Born relation (5. F. Halzen and R.

The other large contribution ∆ρ. which represents the loop (5. m2b ) + m2b F (m2b . It is indeed forbidden in QED and QCD where virtual particle effects are suppressed by “inverse” powers of their masses. ǫ = 0 and diagrams of this type are not possible.113) by the replacement (1 + ∆r) → (1 − ∆r)−1 in (5..5.5. in fact.5. where only equal mass fermions and antifermions appear in neutral photon loops. prohibited in QED by general arguments. m2t ) .114) and (5. |Utb |2 ≃ 1. + (5.5.116) embodies this requirement because ǫ = 0 for photon loops..109).5. (5. see (3. GF ∆ρ ≃ 2 ǫ .112).114) ∆ρ = 4π ω(z − ω) with F (m21 . Its value is given by   z α NC |Utb |2 m2t F (m2t .5. m22 ) = Z 0 1   dx x ln m21 (1 − x) + m22 x . The diagram has the important property that.5. They are.111). (5.116) 3π So in QED.115) where NC = 3 is the number of colors and Utb is the CKM matrix element. the appearance of an m2t /z term is a characteristic feature of the electroweak theory. (5. is our primary focus here.117) ≃ 4π 16π c2 s2 z The appearance of a m2t /z contribution to an observable is far from routine. We already discussed the running of α from the small lepton masses to mZ . defining mt = mb + ǫ. (5. (5.115) in the form   2m2b m2t GF m2t 2 2 mt + mb − 2 ∆ρ = ln 4π mt − m2b m2b 3α 1 m2t GF 2 mt ≃ .can imagine summing the series + . Conversely. ∆ρ provides us with a most fundamental probe of electroweak theory short of discovering the Higgs boson. This can be seen by rewriting (5.5. 171 .5.5.5.

Recall that charged weak currents couple with strength GF . The Higgs particle makes a contribution to ∆r: H m2H 11α 1 ln . (5.110) yields (∆r)calculated = ∆α − c2 ∆ρ = 0.5. cf. consistent with the standard model value (5.64): z ≃ 0.4.5. Using (5. = 48π c2 z ∆h = (5. s2 (5. in fact. a contribution too small to be sensed by the simple analysis presented above.23).5.110) requires a non-vanishing value of ∆ρ.1.3.5. 2 α(mZ ) 137 (5.We are now ready to illustrate that ∆ρ 6= 0 and is.066 .035 .037 .281 GeV)2 We next recall (3.5.119).0086 and (5. We leave it as an exercise to insert errors into the calculation and show that our argument survives a straightforward statistical analysis. Other measurements support the electroweak model’s radiative correction associated with the t¯b loop ∆ρ. ω(z − ω) (5.24) and (5. we obtain that ∆ρ = 0.5.5.59) we obtain that ∆h < 0.111): ∆α ≃ 1 − The crucial point is that ∆rexp 6= ∆α. 172 .5.118) 128 α(0) ≃1− ≃ 0. see (5.5.117) calculated using the experimental value of the mass of the top quark.1.118) and (5.121) W From (5.5.0006.120) in agreement with the experimental value (5.117). Using (5.5. We first determine the experimental value of ∆r from (5. constraints on the value of mH . but definite.119) ∆rexp ≃ 1 − (37.109).5.5. The O(α) standard model relation (5.1.10) and (5. while neutral currents couple as ρGF .118). More sophisticated analyses which include the dominant O(α2 ) corrections are now yielding weak. The quantity ∆r is in principle sensitive to the Higgs mass.

5.5.117). In the end a professional approach follows the technique we previously mentioned: generalize the theoretical expressions for the observables (5. The problem can be quantified by rewriting (5.5.5.The neutral current decay of Z into neutrinos is therefore proportional to ρGF : √ 3 2 3 Γ(Z → ν ν¯) = (ρGF ) m . mW ) plane is shown in Fig.5.122) 24π Z The measured value of 499.5. It will not have escaped the reader’s attention that the precision of the confrontation between theory and experiment is limited by the relatively large errors on the measurements of mW and mt . which depends on the unknown mass of the Higgs boson. Both the direct and the indirect contour curves prefer a low value for the Higgs mass.5 MeV is larger than the value calculated from the above equation which is 497. The combined LEP2 and Tevatron data (solid line) prefers a region outside the diagonal band. (5. successfully testing the standard model at the level of electroweak radiative corrections. of course.0 ± 1. (5. mH ) . The program is however far from complete.5. Also shown is the contour curve corresponding to the direct measurements of both quantities at the Tevatron and LEP2.119) and (5. The diagonal band in the figure shows the constraint between the two masses within the standard model.5.87)–(5. mt . although the statistics are not overwhelming.121). electroweak radiative corrections are evaluated to predict the masses of the top quark and the W -boson. bridging the gap. The radiative corrections predicted by the standard model have successfully confronted experiment.117) increases ρ to a value 1 + ∆ρ = 1.5. and to a small extent also on the hadronic vacuum polarization (small arrow labeled ∆α). 5. Using the Z-pole measurements of SLD and LEP1. require the detection of the Higgs particle within this band.118) and (5. The two contours overlap.110) as ∆rexp = F (mW . the loop contribution (5. The resulting 68% CL contour curve in the (mt . Nevertheless. Confirmation of the standard model will.0086.5. (5.123) using (5.5.9. Failure to do so will undoubtedly raise the question of the precision of the 173 .91) to 1-loop and show that all measurements yield a common value of sin2 θw .

5 LEP2 and Tevatron (prel.4 80. 174 .5: Contour curves of 68% CL in the (mt . The band shows the correlation between mW and mt expected in the standard model.3 ∆α mH [GeV] 114 300 150 1000 175 200 mt [GeV] Figure 5.July 2008 80. mW ) plane for direct measurements and the indirect determinations.) LEP1 and SLD mW [GeV] 68% CL 80.

Phys.5. H. Lett. the deficit comes mainly from Latm ∼ 102 − 104 km.5 − 4σ. S. 26 Y. Rev. Lett. 5656 (2001).25 Data collected by the Sudbury Neutrino Observatory (SNO) in conjuction with data from Super-Kamiokande (SK) show that solar νe′ s convert to νµ or ντ with CL of more than 7σ. Phys. Phys. Phys. see (5. Rev. Ashie et al. 179 175 . most likely converting to ντ . 90. 81. B 539.6. [Super-Kamiokande Collaboration]. 1158 (1998) [Erratum-ibid. 3999 (2000). The evaluation of the threshold contribution of the tt¯ loops to the same integral is not totally understood. Lett. [SuperKamiokande Collaboration].24 The evidence of atmospheric νµ disappearing is now at > 15σ. 93. Fukuda et al.121) with experiment. 112005 (2005). Phys. 1810 (1999). Fukuda et al. 85.Phys. 82. Lett. [Super-Kamiokande Collaboration]. 051801 (2004). Some doubts remain about the accuracy of the e+ e− data in the vicinity of charm thresholds which are used to evaluate the charm quark contribution to the running of α. Fukuda et al. This should be sufficient to confront Higgs vacuum polarization effects such as (5. [Super-Kamiokande Collaboration]. [Super-Kamiokande Collaboration]. 81. Lett.26 The KamLAND 24 Y.6 5.5. Rev. The angular distribution of contained events shows that for Eν ∼ 1 GeV. 86.computations. 041801 (2003). Rev. These most likely represent the true limitation of the calculation but neither problem is likely to preclude the indirect measurement of mH . 178 (2001). State of the art calculations include all dominant 2-loop effects. convincing experimental evidence exists for (time dependent) oscillatory transitions να ⇋ νβ between the different neutrino flavors.1 Lepton Flavor Mixing Neutrino Oscillations At present. D 71. 81. Lett. Rev. Fukuda et al. Lett. B 433. 25 S. B 511.111). 1562 (1998). Y. Phys. The simplest and most direct interpretation of the atmospheric data is that of muon neutrino oscillations. 4279 (1998)]. S. Rev. Rev. Lett. Phys. Phys. Phys. Lett. These results have been confirmed by the KEK-toKamioka (K2K) experiment which observes the disappearance of accelerator νµ ’s at a distance of 250 km and finds a distortion of their energy spectrum with a CL of 2. 5. 9 (1998). Rev. [K2K Collaboration]. Lett. Phys. Ahn et al.

[SNO Collaboration]. JHEP 0709.e.e. Rev. but rather mass eigenstates.6. they have the time dependence |νi (t)i = e−iEi t |νi i with Ei = q p2 + m2i ≈ p + m2 m2i ≈E+ i . [KamLAND Collaboration]. K. 497 (1981). Lett. mixing matrix) |να i = X i Uαi |νi i ⇔ |νi i = X α (U „ )iα |να i = X α ∗ Uαi |να i .6. the states |νi i are stationary states.. Lett. Phys. 176 .124) with U „ U = 1.126) The number of parameters of an n × n unitary matrix is n2 . X ∗ Uαi Uβi = δαβ and i X ∗ Uαi Uαj = δij .28 The flavor eigenstates |να i and the mass eigenstates |νi i are related by a unitary transformation U (i. (5.Collaboration has measured the flux of ν e from distant reactors and find that ν e ’s disappear over distances of about 180 km.6.6. E. Nuovo Cim. (5. 31. 2p 2E (5. For these it is convenient to take the 12 n(n − 1) “weak mixing angles” of an n-dimensional rotation and 1 (n − 1)(n − 2) “CP -violating phases. Ahmed et al.27 All these data suggest that the neutrino eigenstates that travel through space are not the flavor states that we measured through the weak force.. i. 081801 (2005). S. S. N.127) (5. Akhmedov. Phys. 28 Contrariwise. (5. charged leptons are states of definite mass and hence cannot undergo oscillations. It is easy to see that 2n − 1 relative phases of the 2n neutrino states can be redefined such that (n − 1)2 independent parameters are left.. 94. 92. Araki et al.6. 181301 (2004).e. Rev. Pakvasa.128) (2002).125) i ∗ For antineutrinos one has to replace Uαi by Uαi . 27 T. i. i.. Lett.” 2 Being eigenstates of the mass matrix. 116 (2007).e. |¯ να i = X i ∗ Uαi |¯ νi i .

(5.130) with Dij = δij e−iEi t (diagonal matrix).129) |ν(t)i = Uαi e−iEi t |νi i = i i. t) A(ν X ∗ −i(Ei −Ej ) = Uαi Uβi e t i = δαβ + X i = δαβ + X i6=j with  −i(Ei −Ej )t  ∗ Uαi Uβi e −1  −i∆ij  ∗ Uαi Uβi e −1 .6.128) into (5. (5.6. t) = Uαi Uβi e i = X ∗ Uαi Uβi − e im2 iL 2E .130) and extracting an overall phase factor e−iEt X im2 t ∗ − 2Ei A(να → νβ .) Thus a pure flavor state |να i = i Uαi |νi i. (5.6. present at t = 0.β The time dependent transition amplitude for the transition from flavor να to flavor νβ therefore is X ∗ −iEi t A(να → νβ ) ≡ hνβ |ν(t)i = Uαi Uβi e i X = Uαi δij e−iEi t (U „ )jβ i. develops with time into the state X X ∗ −iEi t Uαi Uβi e |νβ i .6. An equivalent expression for the transition amplitude is obtained by inserting (5.133) .131) i where L = ct (recall c=1) is the distance of the detector. from the να source.27 177 δm2ij L E (5. (Here it is assumed that neutrinos P are stable. in which νβ is observed. For an arbitrary chosen fixed j the transition amplitude becomes e α → νβ .j = (UDU „ )αβ . t) = eiEj t A(να → νβ .where E ≈ p is the total neutrino energy. ∆ij = (Ei − Ej ) = 1.6.6.6.132) (5.

i. if time reversal invariance holds. which also follows directly from the CP T theorem: C changes particle into antiparticle. In (5.6.6.6. The number of parameters for the transition amplitude is then 21 (n − 1)(n + 2).6.e.6.29 The transition probabilities are obtained by squaring the moduli of the amplitudes (5. the following relation holds for transformations between neutrinos and antineutrinos.6. one has ˜ α) . by n(n − 1) real parameters.6.126) we obtain the amplitudes for the transitions between antineutrinos X ∗ A(¯ να → ν¯β . The transition amplitudes are thus given by the (n−1)2 independent parameters of the unitary matrix (which determines the sizes of the oscillations) and the n − 1 mass square differences (which determine the frequencies of the oscillations)..6. If CP is conserved in neutrino oscillations.130) . Uαi and Uβi are real in (5. U is an orthogonal matrix (U −1 = U T ) with 12 n(n − 1) parameters. t) = Uαi Uβi e−iEi t ..134) i Therefore. E in GeV and δm2ij = m2i − m2j in eV2 .6.6.130) and (5.132) the unitarity relation (5.e. all CP -violating phases vanish and the Uαi are real. Using (5.134).134).135) If CP is conserved.136) Therefore.6.125) has been used.130) and (5..when L is measured in km. and T reverses the arrow indicating the transition A(¯ να → ν¯β ) = A(νβ → να ) 6= A(να → νβ ) . CP violation can be searched for by e. comparing the oscillations να → νβ and νβ → να . i. P provides the necessary flip from left-handed neutrino to right-handed antineutrino and vice versa. (5. That is. A(¯ να → ν¯β ) = A(να → νβ ) = A(¯ νβ → ν¯α ) = A(νβ Φν (5.g. comparing (5. (5.

2 .

.

.

X .

∗ −iEi t .

Uαi Uβi e Pνα →νβ = .

.

.

.

K. Phys. Phillips. 2084 (1980). 178 . D. Lett.6.137) V. Barger. N. (5. Rev. 45. J. i X ∗ ∗ = δαβ − 4 ℜe (Uαi Uβi Uαj Uβj ) sin2 ∆ij + 2 X i>j 29 i>j ∗ ∗ ℑm (Uαi Uβi Uαj Uβj ) sin 2∆ij . Whisnant and R.

Lett. we can define a mass basis as follows. τ ) are expanded in terms of 3 mass eigenstates |νi i (i = 1.26 . (5.6. With this in mind. the observed up-down asymmetry leads to a mixing angle very close to maximal. δm2atm = 2. sin2 2θatm > 0. 415 (1999).6 eV2 −0.In the standard treatment of neutrino oscillations. T. 3). is sufficient to generate “νµ -ντ interchange symmetry. reactor data suggest |Ue3 | ≪ 1. Petcov. On the other hand.141) is the eigenstate orthogonal to |ν3 i. B 466. which requires δm2atm ∼ 10−4 − 10−2 eV2 . and |ν1 i = sin θ⊙ |ν ⋆ i + cos θ⊙ |νe i . θatm ≃ 45◦ and ℜe(Ue3 ) ≃ 0.140) where θ⊙ is the solar mixing angle and 1 |ν ⋆ i = √ (|νµ i − |ντ i) 2 (5.” To simplify the discussion hereafter we use the fact that |Ue3 |2 is nearly zero to ignore possible CP violation and assume that the elements of U are real. the flavor eigenstates |να i (α = e. 77 (2002). Phys. atmospheric neutrino data suggest that the corresponding oscillation phase must be maximal. Apollonio et al. M. Moreover.6. (5. 2 (5. Phys. Bilenky.142) |ν ⋆ i = sin θ⊙ |ν1 i + cos θ⊙ |ν2 i .139) 1 |ν3 i = √ (|νµ i + |ντ i) . S.6.6.35 −0. assuming that all upgoing νµ ’s which would yield multiGeV events oscillate into a different flavor while none of the downgoing ones do. D.4 × 10 2 30 and tan2 θatm = 1+0.138) |ν2 i = cos θ⊙ |ν ⋆ i − sin θ⊙ |νe i . 2. The combined analysis of atmospheric neutrinos −3 with K2K leads to a best fit-point and 1σ ranges.2+0. Inversion of the neutrino mass-to-flavor mixing matrix leads to |νe i = cos θ⊙ |ν1 i − sin θ⊙ |ν2 i (5. ∆atm ∼ 1. Nicolo and S. In such a case.143) and 30 M. (5. Lett. 179 . [CHOOZ Collaboration]. This twin happenstance.6. B 538. µ.85.6.

144) and by substracting these same equations the ντ eigenstate. For ∆ij ≫ 1 (as would be the case for far-out neutrinos propagating over cosmic distances).6. by adding Eqs. 460.6. i.6.6.6. C.j i=j . Gonzalez-Garcia and M. (5. M. averaging over sin2 ∆ij in (5.6.Finally. The combined analysis of Solar neutrino data and KamLAND data are consistent at the −5 3σ CL. Rept.2+0. Phys.3 eV2 and −0. (5.6.150) For a general discussion of the mixing parameters see e.39+0. 1 (2008).6. 1 |νµ i = √ [|ν3 i + sin θ⊙ |ν1 i + cos θ⊙ |ν2 i] .6. (5.146) cancel each other.6. 31 (5.6.04 .146) i 2 Since δαβ = δαβ . (5.148) (5. (5.j = δαβ − X Uαi Uβi i !2 i + X 2 2 Uαi Uβi . Consequently.147) Uβi P (να → νβ ) = Uαi i The probabilities for flavor oscillation are then P (νµ → νµ ) = P (νµ → ντ ) = 1 (cos4 θ⊙ + sin4 θ⊙ + 1) . yielding X 2 2 .3 × 10 31 tan2 θ⊙ = 0.149) and P (νe → νe ) = cos4 θ⊙ + sin4 θ⊙ .05 −0. Maltoni. 2 (5. (5. 4 P (νµ → νe ) = P (νe → νµ ) = P (νe → ντ ) = sin2 θ⊙ cos2 θ⊙ .6.145) can be re-written as X X P (να → νβ ) = δαβ − Uαi Uβi Uαj Uβj + Uαi Uβi Uαi Uβi 1>j i.137) we obtain X P (να → νβ ) = δαβ − 2 Uαi Uβi Uαj Uβj .145) i>j Now. using 2 P = P − P Eq. 180 .g. the phases will be erased by uncertainties in L and E.. with best-fit point and 1σ ranges: δm2⊙ = 8. the first and second terms in Eq. (5.141) one obtains the νµ flavor eigenstate.140) and (5.

because no righthanded neutrinos exist in the standard model. 33 F. Phys. flavor identification of cosmic neutrinos on a statistical basis becomes possible.33 Altogether. Such a beam from the heavens could be used to study the neutrino oscillation parameters by comparing flavor ratios in the direction of the beam and the rest of the sky.32 Such a unique ratio would appear above the 1:1:1 background in the direction of the neutron source. However. which couple a right-handed fermion with its left-handed doublet and the Higgs field. H. 5. Anchordoqui. In contrast. B 593.151) Pνα detected = Uαj j β Straightforward calculation shows that any initial flavor ratio that contains we = 1/3 will arrive at Earth with equipartition on the three flavors. 42 (2004). Lett. originating via neutron β-decay. (5. J. the prediction for a pure ν¯e source. Science 315. Halzen.6. so that the relative fluxes of each P 2 mass eigenstate are given by wj = α ωα Uαj . neutrinos are massive and therefore the standard model needs to be extended as we discuss next. Goldberg. Weiler. 2. Yukawa interactions (2. 181 . A. has different implications for the flavor ratios: we = 1 yields Earthly ratios ∼ 5 : 2 : 2. their initial flavor ratios of nearly 1 : 2 : 0 should arrive at Earth democratically distributed. With the growth of neutrino observatories.4]. after spontaneous symmetry breaking [see Sec.2 How to kill a vampire In the standard model masses arise from Yukawa interactions. Halzen and T. F. opening up a window for discoveries in particle physics not otherwise accessible to experiment.104) 32 L. let the ratios of neutrino flavors at production in the cosmic sources P be written as we : wµ : wτ with α wα = 1. 66 (2007).Now.4. From our previous discussion. we conclude that the probability of measuring on Earth a flavor α is given by X X 2 2 wβ Uβj . So there is a fairly robust prediction of 1:1:1 flavor ratios for measurements of cosmic neutrinos.6. Since neutrinos from astrophysical sources are expected to arise dominantly from the decay of charged pions (and kaons) and their muon daughters.

1)0 . in order to introduce a neutrino mass term we must either extend the particle content. these “sterile” states would be essentially undetectable and unproduceable. or else abandom gauge invariance and/or renormalizability. and ντ do not have a definite mass. Forcing MN = 0 leads to a Dirac mass term. In this section we illustrate different types of neutrino mass terms. With the particle contents of the standard model and the addition of an arbitrary m number of sterile neutrinos one can construct two types of mass terms that arise from gauge invariant renormalizable operators −LMν = m X X α=e.c. Such a mass term conserves total lepton number.153) similarly to the charged fermion masses. Vν and VνR which perform the biunitary 182 . This is achieved through the transformations νLα = 3 X αk Vν νLk . which is generated after spontaneous electroweak symmetry breaking from Yukawa interactions v Yν iα ν¯Rsi φ¯„ LLα ⇒ MD iα = Yν iα √ . in general. MD is a complex m × 3 matrix. For m = 3 we can identify the three sterile neutrinos with the right-handed component of a four-spinor neutrino field.154) k=1 k=1 with two 3 × 3 unitary matrices. the flavor neutrino fields νe .6. The massive neutrino fields are obtained via diagonalization of LMν . Therefore.152) where ν C indicates a charge conjugated field (ν C = C ν¯T ). and MN is a symmetric matrix of dimension m × m.τ i=1 1 C + h.µ. having no gauge interactions. νµ .6. a complex 3 × 3 matrix.6. . MD iα ν¯Rsi νLα + MN ij ν¯Rsi νRsj 2 (5. assuming we keep the gauge symmetry and we introduce an arbitrary number m of additional right-handed neutrino states νRs (1.leave the neutrinos massless. Since the matrix Y is. but this cannot happen because any neutrino mass term that can be constructed with standard model fields would violate the total lepton symmetry. but it breaks the lepton flavor number symmetries. νRsj = 3 X VνR jk νRk (5. 2 (5. One may wonder if neutrino masses could arise from loop corrections or even by non-perturbative effects.

Therefore. diagonalization „ MD Vν VνR jk v  „ √ = = mk δjk . whithin this set up there is no explanation to the fact that neutrino masses happen to be much lighter than the corresponding charged fermion masses (see Fig. Such a term is different from the Dirac mass term in many important aspects. VνR Yν Vν 2 (5. would have to be introduced to explain the data.156) k=1 k=1 where νk = νLk + νRk are the Dirac fields of massive neutrinos.6). This is the reason that such terms are not allowed for any charged fermions which. because new particles. it can appear as a bare mass term. Furthermore. it breaks lepton number by two units. 183 .c.ν3 e ν2 ud s µ c b τ t ν1 10−4 10−3 10−2 10−1 100 101 102 103 m 104 105 106 107 108 109 1010 1011 1012 [eV] Figure 5. Moreover. 5. since it involves two neutrino fields. neutrino masses receive an important contribution from the Majorana mass term. Note that to get reasonable neutrino masses (below the eV range). It is a singlet of the standard model gauge group. If MN 6= 0. as in this case all particles acquire their mass via the same mechanism. the Yukawa couplings would have to be unnaturally small Yν iα < 10−11 (for charged fermions the Yukawa couplings range from Yt ≃ 1 for the top quark down to Ye ≃ 10−5 for the electron).6. The resulting diagonal mass term can be written as −LMν = 3 X mk ν¯Rk νLk + h.6. carry U(1)EM charges. such a term is allowed only if the neutrinos carry no additive conserved charge. In addition. More generally.155) with real positive masses mk . = 3 X mk ν k νk (5. if MN = 0 the standard model is not a complete low-energy effective theory. such as sterile neutrinos. by definition.6: Order of magnitude of the masses of quarks and leptons.

. . because Yukawa couplings are expected to be unnaturally small. ~νRs ) is a (3+m)-dimensional vector. In general. Vν . Majorana. .k + νmass. since a Dirac mass term is forbidden by the symmetries of the standard model. We have seen that the order of magnitude of the elements of the Dirac mass matrix MD is expected to be smaller than v.159) In terms of the resulting 3 + m mass eigenstates.k = (Vν „ ~ν )k + (Vν „ ~ν )C k .157) (5. 2 k=1 2 k=1 (5.6.6.6.6. and are referred to as Ma34 jorana neutrinos. . 184 . νM = νM .k + ν¯mass. (5.161) C which obey the Majorana condition. Notice that this condition implies that there is only one field which describes both neutrino and antineutrino states.152) can be rewritten as −LMν = 1 C ~ν Mν ~ν + h. ~νmass = (Vν )„ ~ν .In general (5.k νmass. The matrix Mν is complex and symmetric. m2 .6. which are Dirac particles. it can arise only as a consequence of symmetry breaking and hence Dirac mass terms are proportional to the symmetry-breaking scale. Thus a Majorana neutrino can be described by a two-component spinor unlike the charged fermions. (5.6. 171 (1937).160) where C νM k = νmass.k νmass. 14.157) can be rewritten as −LMν 3+m X  1 3+m 1X C C = mk ν¯mass. (5.6. so that Vν T Mν Vν = diag(m1 . 2 where Mν = 0 MDT MD MN ! . and are represented by four-component spinors.158) C T and ~ν = (~νL . It can be diagonalized by a unitary matrix of dimension (3 + m).k = mk ν¯M k νM k . Nuovo Cim. This fact is often summarized by saying that Dirac masses are protected by the symmetries of the standard 34 E. (5. m3+m ) . .c.

since a Majorana mass term is a standard model singlet.6. Thus. This is the see-saw mechanism which explains naturally the smallness of light neutrino masses. The light 3 × 3 mass matrix Ml and the heavy m × m matrix Mh are given by Vν ≃  −MN−1 MD Vl Ml ≃ −VlT MDT MN−1 MD Vl . the seesaw mechanism implies the effective low-energy mixing of three Majorana 35 P. On the other hand.model.162) 0 Mh with   1 − 21 MD„ MN∗ −1 MN−1 MD Vl MD„ MN∗ −1 Vh    1 − 21 MN −1 MD MD„ MN∗ −1 Vh (5. The mass matrix can be diagonalized by blocks. Minkowski. 912 (1980). In this case. Phys.  Mh ≃ VhT MN Vh . 185 . Mohapatra and G. which could be much higher than the scale of electroweak symmetry breaking hφi. the elements of the Majorana mass matrix MN are not protected by the standard model symmetries. however. whereas the light masses are given by the eigenvalues of Ml . whose elements are suppressed with respect to the elements of the Dirac mass matrix MD by the very small matrix factor (MD T MN−1 ). N. It is plausible that the Majorana mass term is generated by new physics beyond the standard model and the right-handed chiral neutrino fields νRs belong to nontrivial multiplets of the symmetries of the high energy theory. Senjanovic. depending on the specific values of the elements of MD and MN . Phys. 44. the three flavor neutrinos are mainly composed by the three light neutrinos. up to corrections of order (MN−1 MD ) ! M 0 l Vν T Mν Vν ≃ (5.164) The heavy masses are given by the eigenvalues of MN . Lett.6.163) where Vl and Vh are 3 × 3 and m × m unitary matrices respectively. Because the off-diagonal block elements of Vν are very small. that the values of the light neutrino masses and their relative sizes can vary over wide ranges. (5. 421 (1977). Lett. the elements of the mass matrix MN are protected by the symmetries of the high energy theory and their order of magnitude corresponds to the breaking scale of these symmetries.35 Note. Rev. R.6. B 67.

As a matter of fact. (5. The answer is off by a factor of 2 because the discovery of parity violation and neutral currents was in the future and introduces an additional factor 1 − m2W /m2Z .    1 πα GF = √ 2 (1 + ∆r) . If one calculates the radiative corrections to the mass µ appearing in the Higgs potential.neutrinos with an approximately unitary 3 × 3 mixing matrix U composed by the first three rows and the first three columns of Vν . There is no feeling though that we are now dotting the i’s and crossing the t’s of a mature theory. the same theory that withstood the onslaught of precision experiments at LEP/SLC and the Tevatron yields a result that grows quadratically.166) . The most evident of unanswered questions is why the weak interactions are weak — in gauge theory the only natural values for mW are zero or the Planck mass. the difference between the bare and renormalized masses is ! X 1 9g 2 + 3g ′2 + 24 λ − 8 Nf Yf2 Λ2 ∆µ2 = 64π 2 f ≃ 3 (2m2W + m2Z + m2H − 4m2t ) Λ2 . Although Fermi adjusted mW to accommodate the strength and range of nuclear radioactive decays. the present victories are bittersweet. 5. Already in 1934 Fermi provided an answer with a theory that prescribed a quantitative relation between the fine structure constant and the weak coupling.7. one can readily obtain a value of mW of 40 GeV from the √ observed decay rate of the muon for which the proportionality factor is π/ 2. 2 2 16π v 186 (5. GF ∼ α/m2W . Besides regular higher order diagrams.7 Closure The saga of the standard model is still exhilarating because it leaves all questions of consequence unanswered.165) 1 − m2W /m2Z 2mW Fermi could certainly not have anticipated that we now have a renormalizable gauge theory that allows us to calculate the radiative correction ∆r to his formula. they have been observed. and the model does not contain the physics that dictates why its actual value is of order 100 GeV.7. loops associated with the top quark and the Higgs boson contribute.

. and Λ is a cutoff. this translates into a dangerous contribution to the Higgs vacuum expectation value which destabilizes the electroweak scale.where g and g ′ are the SU(2)L × U(1)Y gauge couplings. L (mW ) = |µ2 | H „H + λ(H „ H)2 + LSM 4 Λ Λ (5. just like Fermi anticipated particle physics at 100 GeV in 1934. m2Z = 14 (g 2 + g ′2 )v 2 . Following Weinberg. for mH = 115 − 200 GeV. the electroweak gauge theory requires new physics to tame the divergences associated with the Higgs potential. It is generally assumed that this indicates the existence of new physics beyond the standard model. The optimistic interpretation of all this is that. m2H = 2λv 2 . m2t = 21 Yt2 v 2 . . m2W = 14 g 2 v 2 .36 Upon minimization of the potential. The standard model works amazingly well by fixing Λ at the electroweak scale. Avoiding fine tuning requires Λ ∼ CERN LHC. λ is the quartic Higgs coupling. For example.167) where the operators of higher dimension parametrize physics beyond the standard model. Yf are the Yukawa couplings.7. By the most conservative estimates this new physics is within our < 2 − 3 TeV to be revealed by the reach. Nf = 1 (3) for leptons (quarks). 1 1 1 gauge Yukawa + LSM + L 5 + 2 L 6 +. . . v = 246 GeV.

.

.

∆µ2 .

δv 2 .

.

7. (5.168) .

µ2 .

pushing Λ to 10 TeV. where we have implicity used v 2 = −µ2 /λ [valid in the approximation of disregarding terms beyond O(H 4 ) in the Higgs potential]. The data push some of the higher order dimensional operators in Weinberg’s effective Lagrangian to scales beyond 10 TeV. One can even make this stick to all orders and for Λ ≤ 10 TeV. Dark clouds have built up around this sunny horizon because some electroweak precision measurements match the standard model predictions with too high precision. exactly! This has been dubbed the Veltman condition. Polon. Acta Phys. Veltman. possibly even higher. B 12. 187 . can be accommodated by the observations once one eliminates the dominant contribution. The problem is now “solved” because scales as large as 10 TeV. G. 437 (1981). = v 2 ≤ 10 ⇒ Λ = 2 − 3 TeV . 36 M. Some have resorted to rather extreme lengths by proposing that the factor multiplying the unruly quadratic correction (2m2W + m2Z + m2H − 4m2t ) must vanish. this requires that mH ∼ 210 − 225 GeV. J.

non-standard interactions. As for every boson there is a companion fermion. 188 . as it should be. the good news is that all involve new physics in the form of scalars. the bad divergence associated with the Higgs loop is cancelled by a fermion loop with opposite sign.7). leading us to unexpected discoveries. 52 (1974). Zumino. Phys. new gauge bosons. and that all known fermions (bosons) have a supersymmetric boson (fermion) partner. 37 Supersymmetry predicts that interactions exist that would change fermions into bosons and vice versa. Lett. even though higher scales can be rationalized when accommodating selected experiments. B 70. There has been an explosion of creativity to resolve the challange in other ways. This is very unlikely. Let’s contemplate the possibilities.7: Supersymmetry offers a neat solution of the bad behavior of radiative corrections in the standard model. The Veltman condition happens to be satisfied and this would leave particle physics with an ugly fine tuning problem. J. 5. B 49. 39 (1974).H Figure 5. Nucl. it is possible that we may be guessing the future while holding too small a deck of cards and LHC will open a new world that we did not anticipate. LHC must reveal the Higgs physics already observed via radiative correction. . it is already fine-tuned at a scale of ∼ 2 − 3 TeV. It must appear at 2 − 3 TeV. Alternatively. . Minimal supersymmetry is a textbook example.37 Even though it elegantly controls the quadratic divergence by the cancellation of boson and fermion contributions (see Fig. Particle physics would return to its early traditions where experiment leads theory. Wess and B. Phys. and where innovative techniques introduce new accelerators and detection methods that allow us to observe with an open mind and without a plan. or at least discover the physics that implements the Veltman condition.

04 −0. G. and 1 A. [SDSS Collaboration]. J. [Supernova Search Team Collaboration]. Rev. however. [Supernova Cosmology Project Collaboration]. Suppl. Astrophys.138 ± 0. with one concerned with the universe’s elementary constituents and the other concerned with the universe as a whole. establishing a new paradigm of cosmology.012. elementary particles and cosmology seem to be completely different branches of physics. J. M.03 . the Supernova Search Team. unknown dark energy component with strongly negative pressure (ΩΛ = 0.04). Perlmutter et al. 565 (1999).Chapter 6 Dark Side of the Universe At first sight.30 ± 0. Knop et al. Riess et al. Tegmark et al. the matter density Ωm h = 0. 517. 102 (2003). D. Astrophys. 1009 (1998). In recent years. [WMAP Collaboration]. J. The matter budget has only 3 free parameters: the Hub2 ble parameter h = 0. N.1 A surprisingly good fit to the data is provided by a simple geometrically flat universe. the most powerful particle accelerators have recreated conditions that existed in the universe just a fraction of a second after the Big Bang.04) and 70% in the form of a new. 103501 (2004). 116. J.70 ± 0. Phys. [Supernova Cosmology Project Collaboration]. S. D 69. opening a window on the very early history of the universe. 598.70+0. 148. A. 189 . Spergel et al. Astron. a flood of high-quality data from the Supernova Cosmology Project. the Wilkinson Microwave Anisotropy Probe (WMAP). At the same time. Astrophys. in which 30% of the energy density is in the form of non-relativistic matter (Ωm = 0. 175 (2003). R. and the Sloan Digital Sky Survey (SDSS) pin down cosmological parameters to percent level precision.

Ωb h2 = 0.002. M. Nanopoulos. This implies that the structure of the universe is dictated by the physics of as-yet-undiscovered cold dark matter (ΩCDM h2 = 0. M. and their stability often follows as a result of discrete symmetries that are mandatory to make electroweak theory viables (independent of cosmology). Lett.0013 −0. 2 (2005). J. Turner. This is because they combine the virtues of weak scale masses and couplings. They must also be cold or warm to properly seed structure formation. Nucl. 279 (2005). (6. H. 315. 227 (2003). Phys. 552. dn + 3H n = −hσvi(n2 − n2eq ) . Turner. J.5 It is this that we now turn to study. S.0. Ωb h2 = 0. R. Nollett and M. Phys. A. 5 R. S. Olive and M. S. Fields and K. S. Kamionkowski and M. at least on cosmological time scales. Bertone. 50. 1585 (1986) [Erratum-ibid. 3263 (1986)]. based on measurements of deuterium in high redshift absorption systems. Annals Phys. D 34.3 Among the plethora of dark matter candidates. Phys. A. J. 190 .2 the density in baryons. 453 (1984). D 41. Silk. R.020 ± 0. Consider a particle X (of mass mX ) in thermal equilibrium in the early universe. and their interactions must be weak enough to avoid violating current bounds from dark matter searches. Phys. so that they do not disturb the subproccesses of BBN. The particle (or particles) that make up most of the dark matter must be stable.1) dt where n is the present number density of WIMPs. Hooper and J. V. 405. Feng. J. WIMPs –weakly interacting massive particles– represent a particularly attractive and well motivated class of possibilities. 3 G. D 33. Phys. Burles. J. Hagelin. Rev. Goldberg.012) and the galaxies we see today are the remnants of relatively small overdensities in the nearly uniform distribution of matter in the very early universe. Generic WIMPs were once in thermal equilibrium. H is the expansion rate 2 The latter is consistent with the estimate from Big Bang nucleosynthesis (BBN). WIMPs are naturally produced with the cosmological densities required of dark matter. K. Ellis. Srednicki. 3565 (1990). D.4 Moreover. Lett. L1 (2001). L. S. 4 H. K. Griest. D. and non-baryonic. 1419 (1983). Rev. Scherrer and M. B.115 ± 0. B 567. Rev. K.0230+0. D. but decoupled while strongly non-relativistic. Olive. Phys. Astrophys. Cyburt. Turner.0012 . B 238. The evolution of the number density as the universe expands is driven by Boltzman’s equation. Rept.

one obtains xFO ≃ 20 − 30. After freeze out. Note that in the very early universe. with the velocity implied. (6. the number density of a non-relativistic X particle is neq = g  mX T 2π 3/2 e−mX /T (6.2) where g is the number of internal degrees of freedom of the WIMP particle. The product σv is usually referred to as the annihilation cross section. and the X particles freeze out of thermal equilibrium.of the universe at temperature T .0. The freeze out temperature is given by −1  r 45 g mX MPl (a + 6b/xFO ) 1 TFO . H ≥ nhσvi. when n ≃ neq .4) ≡ ≃ ln √ mX xFO 8 2π 3 g⋆ xFO where g⋆ is the number of relativistic degrees of freedom at temperature TFO . for weak scale cross sections and masses. Finally. which depend on the particle mass. For heavy states.0. √ MPl g⋆ (a + 3b/xFO ) 191 (6. however.0. the right hand side of Eq.3) To estimate the relic density. and hσvi is the thermally averaged annihilation cross section of the X particles multiplied by their relative velocity. the rate of depletion due to expansion becomes greater than the annihilation rate. When solved by integration.0. As the temperature falls below mX . when the number density falls enough. the equilibrium number density becomes suppressed and the annihilation rate increases. neq is the equilibrium number density. the density of X particles that remain is given by ΩX h2 ≃ 109 GeV−1 xFO 1 .5) . one is thus left with the calculation of the annihilation cross section (in all the possible channels) and the extraction of the parameters a and b. rapidly reducing the number density. At equilibrium. (6. (6.1) is small and the evolution of the density is dominated by Hubble expansion. we can approximate hσvi with the non-relativistic expansion in powers of v 2 hσvi = a + bhv 2 i + O(hv 4 i) ≃ a + 6b/x .0.

Should we be so lucky. a + 3b/20 (6. anti-matter. such as the LHC. and neutrinos. In addition.0. Freeze out temperatures between about 5 and 80 GeV. leading to an apparent energy imbalance. its contents.6) Thus we see that the observed cold dark matter density (ΩCDM h2 ≃ 0.1 × 3 × 10−26 cm3 /s . but is not expected to significantly effect the result. Many approaches have been developed to attempt to detect dark matter. and its evolution. correspond to WIMPs with 100 GeV < mX < 1500 GeV. Monojets and final states with multiple jets plus E smoking guns for the dark matter hunters. Adding up the standard model degrees of freedom lighter than 80 GeV leads to g⋆ = 92.1) can be obtained for a thermal relic with weak scale interactions. WIMPs would escape the detector without interactions. or very light WIMP. (For a very heavy. the coming years of exploration will not only provide our first incisive probe of the Fermi scale. but they will doubtless open a wondrous new view of the cosmos. may have enough energy to directly produce WIMPs. Once produced. this number may change somewhat. this expression yields ΩX h2 ∼ 0. 192 . or “missing energy” /T could become signature.) Numerically. particle accelerators of the next generation. Such endeavors include direct detection experiments which hope to observe the scattering of dark matter particles with the target material of the detector and indirect detection experiments which are designed to search for the products of WIMP annihilation into gamma-rays. to expose the identity of dark matter it is necessary to measure its non-gravitational couplings. Of course.where a and b are expressed in GeV−2 .

0.0.Appendix A Decay Rate in Terms of M In nonrelativistic quantum mechanics. that is ψ(t) ∼ e−iM t e−Γt/2 .1) where τ ≡ 1/Γ is called the lifetime of the state. (A. the time dependence of ψ(t) for an unstable state must include the decay factor Γ/2. (A. (The particle half-life is τ ln 2. E − M + (iΓ/2) (A.3) ∼ 1 .) Thus. (A.2) where M is the rest mass energy of the state. Such an unstable particle decays according to the exponential law.5) .0. the state is described by the Fourier transform Z χ(E) = ψ(t)eiEt dt (A.0. |ψ(t)|2 = |ψ(0)|2 e−Γt . an unstable atomic state shows up in scattering experiments as a resonance.0.4) the experimenter thus sees a reaction rate of the form |χ(E)|2 ∝ 1 (E − M)2 193 + (Γ/2)2 . As a function of the center-ofmass energy E of the system.

E. they are identified by tracking their decay products.6) The Breit-Wigner formula (A. For example. Particles that decay by strong interactions do not live long enough to leave tracks in an experimentalist’s detector.This function has a sharp peak centered at M with a width determined by Γ. it gives the scattering amplitude for processes in which initial particles combine to form an unstable particle. Long and D. Phys. with Γ ∼ 100 MeV. In 1952.20). d3 pp of the final state particles is dΓ = d 3 pp 1 d 3 pπ + 2 (2π)4 δ (4) (p++ |M| ∆ − pπ + − pp ) . H. in πp scattering. The unstable particle viewed as an excited state of the vacuum. A. The formula has the form of (3. Rev. the ∆++ is formed and rapidly (τ ∼ 10−23 s) decays.0. Nagle. Anderson. E.4) also applies in relativistic quantum mechanics.0. Due to its short lifetime. 85.0. E. 49. Rather. Rev. The mass of the decaying particle is the total energy of these products as measured in its rest frame. The decay rate of the ∆++ (assumed to be at rest) is Number of decays per unit time . Breit and E. the uncertainty in its mass (∼ ~/∆t) is sufficiently large to be directly observable. Phys. 936 (1952). L. which then decays. is a direct analogue of the unstable non-relativistic atomic state. 194 . 2 2 (E − M) + (Γ/2) Γ Γ (A. Fermi. 519 (1936).7) Γ≡ Number of ∆++ particles present Hence. using a beam of π + with varying amounts of energy directed through a hydrogen target (protons). Fermi found that the number of interactions (π + scattered) when plotted versus the pion kinetic energy has a prominent peak around 200 MeV. (A.5) becomes |χ(E)|2 ∝ (Γ/2π) 2π 2π = δ(E − M) . (A.0. πp → ∆++ → π + p+ .8) ++ (2π)3 2Eπ+ (2π)3 2Ep 2E∆ ++ where 2E∆ is the number of decaying particles per unit volume and M is the invariant amplitude which has been computed from the relevant Feynman diagram.0. the differential rate for the decay ∆++ → π + p into momentum elements d3 pπ+ .1. In the narrow-width-approximation (A.2 1 2 G.1 In particular. Wigner.

195 . Therefore. 6 an γ 5 γ 5 ). . . b. 6 an ) = 0. γ 5 } = 0 leads to: (−1)n Tr(γ 5 6 a1 . 6 an ) = Tr(6 a1 . .5. Tr(6 a1 . . . 6 an γ 5 ) = (−1)n Tr(6 a1 . Tr(6 a1 . . now. if n is odd. The trace theorems are (using again the notation 6 a = γµ aµ ): ˆ Tr 1I = 4 ˆ Trace of an odd number of γµ ’s vanishes. Tr(γ µ ) = Tr(γ 5 γ 5 γ µ ) = −Tr(γ 5 γ µ γ 5 = −Tr(γ 5 γ 5 γ µ ) because (γ 5 )2 = 1I because using cyclic property of trace µ = −Tr(γ ) . .Appendix B Trace Theorems and Properties of γ-Matrices Using (1. . 6 an ). The trace of one γ matrix is easy. . the anticonmmutation relation {γ µ . ˆ Tr(6 a 6 b) = 4 a .42) the trace of a product of γ-matrices can be evaluated without ever explicitly calculating a matrix product. b Tr(6 a 6 b) = 21 Tr(6 a 6 b+ 6 b 6 a) = 12 2g µν aµ bν Tr(1I) = 4 a . .

c) − Tr(6 b 6 c 6 d 6 a) . c)(b . •Tr(γ5 6 a 6 b 6 c 6 d) = 4i ǫµνλσ aµ bν cλ dσ .2) where ǫµνλσ = +1 (−1) for µ.0. 3. d) − 8(a . c)(b . ˆ Tr(γ5 ) = 0 Tr(γ5 ) = Tr(γ 0 γ 0 γ 5 ) = −Tr(γ 0 γ 5 γ 0 ) = −Tr(γ 0 γ 0 γ 5 ) = −Tr(γ 5 ).0. and so it must be proportional to ǫµνλσ . d)(b . b)(c . c)]. The overall constant can be easily obtained by plugging in (µνλσ) = (0123). bTr(6 c 6 d) − Tr(6 b 6 a 6 c 6 d) = 8(a . Expressions resulting from the use of the last formula can be simplified by means of the identities: ǫαβγδ ǫαβγδ = −24. d) + (a . (B. b)(c . 1. 2. Interchanging any two of the indices simply changes the sign of the trace. c)] Tr(6 a 6 b 6 c 6 d) = Tr[(− 6 b 6 a + 2a . Tr(γ5 6 a 6 b) = 2i[−g 0µ Tr(γ 1 γ 2 γ 3 γ ν ) + g 0ν Tr(γ 1 γ 2 γ 3 γ µ )]aµ bµ = 0. d) − (a . σ and even (odd) permutation of 0. (B. d) + (a . •Tr(γ5 6 a 6 b) = 0 (B. ν. ǫαµβν ǫρµσν = ǫµναβ ǫµνρσ = −2(δ αρ δ βσ − δ βρ δ ασ ) .ˆ Tr(6 a 6 b 6 c 6 d) = 4[(a . b)(c . c)(b . d)(b . d)(b . and 0 if two indices are the same.1) Tr(γ5 6 a 6 b) = Tr(iγ 0 γ 1 γ 2 γ 3 γ µ γ ν ) aµ bν = [−2ig 0µ Tr(γ 1 γ 2 γ 3 γ ν ) + 2ig 0ν Tr(γ 1 γ 2 γ 3 γ µ ) − Tr(iγ5 γ 1 γ 2 γ 3 γ µ γ ν )] aµ bµ Hence. d) − (a . b)(c .3) Other useful results for simplifying trace calculations (that follow directly from the trace theorems) are: ˆ γµ γ µ = 4 × 1I = 4 196 . d) + 8(a . Hence. b) 6 c 6 d] = 2a . Tr(6 a 6 b 6 c 6 d) = 4[(a . λ. cTr(6 d 6 b) + Tr(6 b 6 c 6 a 6 d) = 8(a . d) − 2a .0. ǫαβγµ ǫαβγν = −6δ µν .

The following relations are useful for the computation of the invariant amplitude of weak interaction processes: •Tr(γ µ p/1 γ ν p/2 ) = 2pµ1 Tr(γ ν p/2 ) − Tr(p/1 γ µ γ ν p/2 ) = 2pµ1 Tr(γ ν p/2 ) − 2g µν Tr(p/1 p/2 ) + Tr(p/1 γ ν γ µ p/2 ) = 2pµ1 Tr(γ ν p/2 ) − 2g µν Tr(p/1 p/2 ) +2pµ2 Tr(p/1 γ ν ) −Tr(p/1 γ ν p/2 γ µ )   = 2 pµ1 Tr(γ ν p/2 + pµ2 Tr(γ ν p/1 ) − g µν Tr(p/1 p/2 ) −Tr(γ µ p/1 γ ν p/2 ) = 4 [pµ1 pν2 + pµ2 pν1 − g µν (p1 .1) and (B. p4 ) − (p1 .0. (B. p4 ) + (p2 . b − 4 6 a 6 b + 4 6 a 6 b = 4a . b ˆ γµ 6 a 6 b 6 cγ µ = (2aµ − 6 aγµ ) 6 b 6 cγ µ = 2 6 b 6 c 6 d − 4(a . p2 )(p3 . p4 )(p2 .8) 197 . p2 )(p3 .5) and because {γ µ .4) •Tr[γ µ (1 − γ 5 )p/1 γ ν (1 − γ 5 )p/2 ] = Tr[γ µ p/1 γ ν p/2 + γ µ γ 5 p/1 γ ν γ 5 p/2 ] − Tr[γ µ p/1 γ ν γ 5 p/2 + γ µ γ 5 p/1 γ ν p/2 ] .0. γ 5 } = 0 we have Tr[γ µ (1 − γ 5 )p/1 γ ν (1 − γ 5 )p/2 ] = 2Tr[γ µ p/1 γ ν p/2 ] + 2Tr[γ 5 γ µ p/1 γ ν p/2 ] .7) •Tr(γ µ p/1 γ ν p/2 )Tr(γµ p/3 γν p/4 ) = 16[pµ1 pν2 + pµ2 pν1 − g µν (p1 . p2 )] . p4 )(p1 . p2 )(p3 . (B. p4 ) + (p1 . p4 )(p2 . we obtain Tr[γ µ (1 − γ 5 )p/1 γ ν (1 − γ 5 )p/2 ] = 2Tr(γ µ p/1 γ ν p/2 ) + 8iǫµανβ p1α p2β . p4 ) + (p2 . p3 )(p2 . c)− 6 c 6 b] 6 a − 4(b . p3 ) − (p1 .0. p3 ) − (p1 .0. p4 ) + 4(p1 . c) 6 a = −2 6 c 6 b 6 a. p2 )] × [p3µ p4ν + p4µ p3ν − gµν (p3 . (B. p4 )] = 16[(p1 . p3 )] .0. p4 ) − (p1 .ˆ γµ 6 aγ µ = γµ (2aµ − γ µ 6 a) = 2 6 a − 4 6 a = −2 6 a ˆ γµ 6 a 6 bγ µ = (2aµ − 6 aγµ )(2bµ − γ µ 6 b) = 4a . p3 )(p1 . p4 )] = 32[(p1 . p3 )(p2 . (B. (B.0. p4 ) + (p1 .0. p2 )(p3 .2) in the second term.6) Using (B. p2 )(p3 . c) 6 a = 2[2(b .

p4 )] = 256(p1 .0. p4 )(p2 . p4 ) + (p2 . p4 ) − 2(p2 . •∐ = Tr[γ µ (1 − γ 5 )p/1 γ ν (1 − γ 5 )p/2 ]Tr[γµ (1 − γ 5 )p/3 γν (1 − γ 5 )p/4 ] = 64[pµ1 pν2 + pµ2 pν1 − g µν (p1 . p3 ) − (p1 .•Tr(γ µ p/1 γ ν γ 5 p/2 )Tr(γµ p/3 γν γ 5 p/4 ) = Tr(γ 5 p/2 γ µ p/1 γ ν )Tr(γ 5 p/4 γ µ p/3 γν ) = (4i)2 ǫαµβν p2α p1β ǫρµσν pρ4 pσ3 = 32[(p1 p3 )(p2 . p3 ) + 2(p1 . p3 ) + 2(δρα δσβ − δρβ δσα )p1α p2β pρ3 pσ4 ] = 64[2(p1 . p4 ) − ǫµναβ ǫµνρσ p1α p2β pρ3 pσ4 ] = 64[2(p1 .0. p2 ) − (p1 . p3 )(p2 .0. p4 ) + (p1 . p3 ) − (p1 . (B. p4 )(p2 .10) 198 . p2 ) + iǫµανβ p1α p2β ] × [p3µ p4ν + p4µ p3ν − gµν (p3 . p4 ) .3). p3 )(p2 . p3 )(p2 . p4 ) − (p1 . p4 ) + 2(p1 . p4 ) + 2(p1 . (B. p3 )(p2 . p4 )(p1 . p4 )(p2 . and to obtain the third line (B.2).9) where to obtain the second line we have used (B. p4 )(p1 . p4 ) + (p2 . p2 )(p3 . p4 ) + iǫµρνσ pρ3 pσ4 ] = 64[(p1 . p4 ) − (p3 .0. p4 )(p2 . p2 )(p3 .0. p2 )(p3 . p4 ) + 4(p1 . p3 )(p1 . p2 )(p3 .1) and (B. p3 ) . p3 )(p2 . p3 )(p1 .

The answer is given by (3. . 199 ..5.5..100). .5.101) 1 Here. (C. . formally introduced in (3. Nowadays we avoid the explicit introduction of a cutoff such as Λ in (3. (3. where the dashed lines represent a test charge “measuring” the electron charge on the left. be absorbed in a redefinition of the bare charge.2) How to compute b. This can be visualized in terms of Feynman diagrams e = e0 + e0 + e0 + . The UV cutoff Λ removes the infinite part of the loop which can. The measured charged is obtained through a perturbative calculation including all vacuum polarization loops. 1 + Π(q 2 ) (C.108).110).1 The latter becomes a parameter to be fixed by experiment. bare refers to the fact that the vertex is stripped of all loops.107) and (3. or Π(q 2 ) is clear. This is standard old-fashioned QED.0.1) The geometric series can be summed to give α(q 2 ) = α0 .Appendix C Dimensional Regularization In QFT a charge is surrounded by virtual e+ e− pairs (vacuum polarization) which are the origin of the s-dependence of α.0. in a renormalizable gauge theory. .5. e2 (q 2 ) = e20 − e20 Π(q 2 ) + e20 Π2 (q 2 ) − .5.

3) where me is the electron mass and k the 4-momentum circulating in the loop.2 The basic idea is to carry out loop integrations in a space with dimensions n < 4. The result is then analytically continued to n = 4 where the UV divergent part of the loop appears as a 1/(n − 4) pole. Nucl.0. where they are finite. this last relation follows by executing the following steps: i) use the Feynman trick for combining denominators 1 = ab Z 0 1 dx  1 ax + b(1 − x) this equation can be verified by direct calculation Z 0 1 2 . Phys. J. Phys. The only modification is the introduction of the ’t Hooft mass µ introduced as a factor µ(4−n) in order to keep the coupling constant dimensionless.4)  1 dx 1 1 = − [x(a − b) + b]2 a − b x(a − b) + b 0   1 1 1 . G.100) iΠµν (q) = −i(q 2 gµν − qµ qν )Π(q 2 )   Z   4−n 2 γµ (k/ + me ) γν (q/ + k/ + me ) dn k Tr (−1) = e0 µ 2 (2π)n [k 2 − m2e ] [(q + k)2 − m2e ] Z 1 Z dn Q = −4e20 µn−4 dx (2π)n  2−n 2 0  gµν n Q + m2e + q 2 x(1 − x) − 2qµ qν x(1 − x) ×    2 Q2 − m2e + q 2 x(x − 1) (C. ’t Hooft. 189 (1972).g. Giambiagi. Nuovo Cim. C. Veltman. for the loop (3. In the last line we have omitted terms linear in Q in the numerator which do not contribute to the integral. − = − a−b a b 2 G.. 455 (1973). 200 . Nucl.which spoils the gauge invariance of the calculation. B 12. G. (C. B 44.0. G. e. Propagators and interaction vertices remain unchanged. 20 (1972). B 61. One instead uses dimensional regularization to compute Π(q 2 ). J.5. ’t Hooft and M. Bollini and J.

0.) . 2 n 4 2  n−4 ln C + · · · . γµ γα γ µ = (2 − n)γα . but notice that γµ γ µ = n 1I .0.8) around n = 4 using the following relations: µ(4−n) = 1 +  n−4 ln 4π + · · · .ii) change the integration variable k by the variable Q = k + qx (C. 2 We thus obtain the desired separation of the n = 4 infinite and finite parts of Π(q 2 ) with  Z 1 2 α 2 − − γE + ln 4π − 6 dx x(1 − x) Π(q ) = 3π n−4 0  2   me + q 2 x(1 − x) × ln + O(n − 4) .10) µ2 C ( 2 −2) = 1 + n 201 . n Γ 2 − (2π)n (Q2 − C)2 2 (16π 2 ) 4 (C.0.7) From (C. (16π ) = 16π 1 + 2  2 n =− Γ 2− − γE (= 0.5772 . (C.0. iii) do the traces as usual. (C. .0.0.6) (C.3) we then find that Z  2  n2 −2  n 8e20 µ(4−n) 1 2 2 Γ 2− Π(q ) = (C.8) dx x(1 − x) me + q x(x − 1) n 2 (16π 2 ) 4 0 by using the relation Z  i n  ( n2 −2) 1 dn Q C = .9) We now make a Taylor expansion of (C.0. 2 n−4 2 and 4−n ln µ2 + · · · .0.5) (this is chosen such that the term in the denominator linear in the integration variable disappears). .

0. W.11) with Π(0) given by (C.5. α(µ) ≡ e2 (µ)/(4π) is related to α by  2 1 1 1 µ . Buras. previously introduced.e.0.14) α(µ2) α 3π f m2f where the sum is over all fermions with charge Qf . (C.101) in the limit of large (−q 2 ). Duke and T. (C. In old-fashioned QED the renormalized charge (i.5. 3 W. A. 202 . Bardeen.110).0.0.13) b= 3π If µ2 is such that other loops of leptons and quarks contribute then ! 1 1 1 X 2 µ2 Qf ln = − .12) just evolves the MS charge from Q2 = 0 to Q2 = µ2 and one sees that 1 . J.. Eq.10).0. (C. D 18. (C. Muta.0.12) implements the so-called modified minimal subtraction γ (MS) renormalization scheme where the terms 2E − 12 ln 4π are subtracted out along with the (n − 4)−1 pole into the renormalized charge. the Thomson charge at q 2 = 0) would be defined as e20 µn−4 .which yields (3. A. e ≡ 1 + Π(0) 2 e2 α≡ 4π (C. vacuum polarization effects are completely absorbed in the “running” renormalized coupling by allowing µ to vary.3 We have now succeeded in computing b appearing in the formal relation (3.0. 3998 (1978).12) − = − ln 2 α(µ ) α 3π m2e Equation (C. Phys. In the modern approach. Rev. D.

0. and they interact via their magnetic moment as well as by their charge. 4. The general expression for the transition amplitude 1 Mott scattering is also referred to as spin-coupling in elastic Coulomb scattering. ~0). a number of other effects become significant. (D. and the scattering behavior diverges from the Rutherford formula.0.2) The Feynamn diagram for scattering of an electron by an external field is shown in Fig. In the so-called “Mott scattering.” the magnetic moment and recoil are taken into account.Appendix D Mott Scattering The scattering of electrons from nuclei has given us the most precise information about nuclear size and charge distribution. The electron is a better nuclear probe than the alpha particles of Rutherford scattering because it is a point particle and can penetrate the nucleus. the electron scattering can be described by the Rutherford formula.1 The electromagnetic field due to −Zeρ(x) may be described as an external field Aµ = (φ. (D. For low energies and under conditions where the electron does not penetrate the nucleus.1.4.1) where using (3. 203 .65) ∇2 φ = Ze ρ(~x) . The probing electrons are relativistic. As the energy of the electrons is raised enough to make them an effective nuclear probe. they produce significant nuclear recoil. because it is mostly used to measure the spin polarization of an electron beam scattering off the Coulomb field of heavy atoms.

x~ φ(~x) d3 x .0.30) and (3. we first integrate by parts Z Z i~ q.31) Z Tf i = (−i) d4 x e jµf i (x) Aµ (x) Z = (−i) d4 xeuf γµ ui ei(pf −pi ) .16) the differential cross section can be written as dσ = |Tf i |2 /T V (number of final states) . Ei (Ef ).7) Now.6) and after that we substitute (D. (initial flux) (D.5) Z Z Ze i~ q.6) into the scattering amplitude (D.0.2.4) Tf i = 2πiδ(Ef − Ei ) Ze2 F (~q) (uf γ0 ui ) .follows from (3.~ x 2 3 2 ei~q .0.0. from (3. Considering the boundary condition.1. (D. (initial flux) = v 204 2E1 V (D.~ x 3 e φ(~x) d x = − 2 ei~q . We write the momentum and energy of the incoming (outgoing) electron as ~ki (~kf ). ~x φ(~x) . φ(~x) → 0 when |~x| → ∞. ~x ρ(~x) d3 x |q| Ze = − 2 F (~q) .x0 euf γ0 ui d3 x ei~q .0. then for k = |~ki | = |~kf |.9) .0. x Aµ (x) .0.2.0.1) Z Z 0 i(Ef −Ei ) .0. (D.0.2) into (D. Tf i = (−i) dx e (D.8) where T and V are the time of the interaction and the normalized volume.0. |q| (D.5) e ∇ φ(~x) d x = −|q| then we substitute (D. |q|2 (D.3) or using (D.4) where ~q = ~pf − p~i .0. ~x φ(~x) Z = (−2πi)δ(Ef − Ei ) euf γ0 ui d3 x ei~q .

Using the above formulae we take over (D.and (number of final states) = d3~kf .0.11) summing final.0. and averaging initial.s f = 2[2E 2 − E 2 + k 2 cos θ + E 2 − k 2 ]   2 2 1 − cos θ = 4 E −k 2   k2 2 2 θ = 4E 1 − 2 sin E 2   2 2 2 θ (D. (D. To obtain the unpolarized cross section.14) |uf γ0 ui | (2π) dσ = 2 s .15) = 4E 1 − v sin 2 205 .0. (D. (D. (2π)3 2EF (D. electron spins   2  2 1 kEdΩ Ze F (~q) 1X 2 .13) where Ei = Ef = E and ki = kf = k.0.0.0.0.7) one delta function remains and 2πδ(Ef − Ei ) = Z T /2 ei(Ef −Ei )t dt = T.10) where v = ki /Ei is the velocity of the incoming electron. (D.11) dσ = T v2Ei (2π)3 2Ef On squaring (D. k − m )g ] 2 si.0.8) to arrive at   1 d3 kf |Tf i |2 .s |~q|2 (2π)3 2E v2E i f where 1X 0 ′0 ′0 0 ′ 2 00 |uf γ0 ui |2 ≡ L00 (e) = 2 [k k + k k − (k .12) −T /2 The remaining delta function can be integrated as follows d3 kf δ(Ef − Ei ) = kf2 dkf dΩδ(Ef − Ei ) = kf Ef dEf dΩδ(Ef − Ei ) = kEdΩ . we rewrite (D.0.

We can now rewrite (D.0.14) using |~q|2 = |~ki − ~kf | = 4k 2 sin2 (θ/2)    2 2 dσ eZ E2 2 2 θ 1 − v sin |F (~q)|2 = 4 4 dΩ 2 4π 4k sin (θ/2) or equivalently .and θ is the angle through which the electron is scattered.

.

  dσ .

.

(Zα)2 E 2 dσ .

.

2 2 θ . ≡ = 1 − v sin dΩ .

point dΩ .

0.0. Putting all this together yields the advertised result .Mott 4k 4 sin4 (θ/2) 2 (D.16) (D.17) where α = e2 /4π.

dσ dσ .

.

0. |F (~q)|2 (D.18) = dΩ dΩ .

206 .Mott with the form factor given by (4.1.3).

0. To obtain the last line. p)(k .. the frame where the initial muon is at rest.e.0.0. To this end. k 2 = k 2 ≃ 0 and q 2 ≃ −2k . p) + 2(k .Appendix E Laboratory Kinematics In this appendix we determine the e− µ− → e− µ− cross section in the laboratory frame.2) 2M E E cos − sin2 = 4 2 q 2 2M 2 where to reach the last line we have used the following kinemtic relations q 2 ≃ −2k . ~0). k ′ . p )(k .1) in the lab frame we find   4 1 1 8e 2 ′ ′ 2 2 2 − q M(E − E ) + 2EE M + M q |M|2 = q4 2 2   4 2 2 8e q q M(E − E ′ ) 2 ′ = 2M E E 1 + − q4 4EE ′ 2M 2 2EE ′   q2 θ 8e4 2 ′ 2 θ .3) .0. k] 4 q  8e4  1 2 ′ ′ 2 2 1 − q (k . and q = k − k ′ . p)(k . (E. (E. i. we return to the exact formula (3. p) + (k ′ . we have used ′ p′ = k − k ′ + p.44) for e− (k) µ− (p) → e− (k ′ ) µ− (p′ ) and neglect only the terms involving the electron mass me 8e4 ′ ′ [(k .1) = 2 2 q4 |M|2 = where mµ ≡ M. 207 (E. k ′ ≃ −2EE ′ (1 − cos θ) = −4EE ′ sin2 (θ/2) . Evaluating (E. p = (M. p) + M q . We want to evaluate the cross section in the lab frame.2. p − k . p′ ) − M 2 k ′ .

8) where the step function Θ(x) is 1 if x > 0 and 0 otherwise.0.0.5) The flux is the product of beam and target densities (2E)(2M) multiplied by the relative velocity which is 1 (i.e.4) To calculate the cross section. To obtain the last line we have used p2 = M 2 and the (E. Now.4).7) and so Z 3 ′ Z 3 ′ dp ′ d p (4) ′ δ (p + q − p ) = dp0 Θ(p′0 ) 2p′0 δ(p′2 − M 2 ) δ (4) (p + q − p′ ) ′ ′ 2p0 2p0 Z = d3 p′ dp′0 Θ(p′0 ) δ(p′2 − M 2 ) δ (4) (p + q − p′ )  = δ (p + q)2 − M 2 = δ(p2 − M 2 + 2p .6) we obtain the relation Z dp′0 2p′0 Θ(p′0 ) δ(p′2 − M 2 ) = 1 (E. we make use of (3. q = −2νM so ν ≡ E − E′ = − q2 . the speed of light) in the limit where me has been neglected. δ ν+ 2M 2M (E.0.0..0. q + q 2 )   1 q2 = .1. 2M (E.0. Substitution of the kinematic 208 .20) dσ = = |M|2 d3 k ′ d3 p′ (4) 1 δ (p + k − p′ − k ′ ) (2E)(2M) 4π 2 2E ′ 2p′0 d3 p′ (4) 1 |M|2 1 ′ ′ E dE dΩ δ (p + q − p′ ) . squaring q + p = p′ we obtain q 2 = −2p . from δ(p′2 − M 2 ) = δ(p′2 p ′2 − M 2 ) 0 −~ p p 1 [δ(p′0 − p~ ′2 + M 2 ) + δ(p′0 + p~ ′2 + M 2 ) = p 2 p~ ′2 + M 2 (E.In addition. 4ME 4π 2 2 2p′0 (E.

= cos − sin dE ′ dΩ q4 2 2M 2 2 2M (E.10) Inserting (E.0.0.9) δ E − 2MA A where A=1+ θ 2E sin2 .5) and using (E. we obtain     dσ q2 (2αE ′ )2 q2 2 θ 2 θ δ ν+ .0.0.0.9) we can perform the dE ′ integration and.0.11) Using (E.8) leads to Z   d3 p′ (4) 1 2EE ′ sin2 (θ/2) ′ ′ δ (p + q − p ) = δ E−E − 2p′0 2M M     2 2E sin (θ/2) 1 ′ −E δ E 1+ = 2M M   1 E ′ = (E.0.3). M 2 (E.0.2) into (E.3) into (E.relation (E. we finally arrive at the following formula for the differential cross section for e− µ− scattering in the lab frame .0.8).0. replacing q 2 by (E.

 ′    dσ .

.

E α2 q2 2 θ 2 θ = cos − . sin dΩ .

using (3.37) with Lµν replaced by (p + p′ )µ (p + p′ )ν we obtain the amplitude for elastic scattering of unpolarized electrons from spinless point-like particles |M|2 = = e4 X [u(k ′ )γ µ u(k)] [u(k ′ )γ ν u(k)]∗ (p + p′ )µ (p + p′ )ν 2 q 4 spins e2 Tr(6 k ′ γ µ 6 kγ ν )(p + p′ )µ (p + p′ )ν .12) (µ) Next.0. we neglect once more the mass of the electron and M again 209 .13) In what follows.lab 2 2M 2 2 4E 2 sin4 (θ/2) E (E. 2q 4 (E.2.0.

p) − (k ′ . k ′ ) − (k .5).0. p)(k ′ . k ′ )2 4 q  (k . p) + 2[(k .denotes the target mass. k ′ ) 2 2 [4M − q ] − 2   2  q4 q2 −q 4 e4 ′ 2 ′ 2 2 − 4EE M + 2M(E − E ) + (4M − q ) = q4 2 4 4  2   4 4 q −q 4e − 4EE ′ M 2 − q 2 + M 2q2 = 4 q 2 2 4 e4 = [4EE ′ M 2 − 4EE ′ M 2 sin2 (θ/2)] 4 q   4 e4 2 ′ 2 (4M EE ) 1 − sin (θ/2) = q4 4 e4 = (4M 2 EE ′ ) cos2 (θ/2) .0. p)](k .14) 4 q After substituting (E. (E. straightforward integration leads to .0. using p + k = p′ + k ′ we obtain  4 4 e |M|2 = 4(k .14) into (E.

 ′  E θ dσ .

.

(E.15) = 4 .0. α2 cos2 .

210 . we see that the sin2 (θ/2) in (E.12) is due to the scattering from the magnetic moment of the muon.15) with the cross section for e− µ− → e− µ− .0. 2 dΩ lab 2 4E sin (θ/2) E Comparing (E.0.

1 The differential cross section follows from the relevant expression of conventional QED in Table 3. It has become conventional to call CF ≡ 12 |c1 c2 | the color factor (although. Similarly. supplemented by the appropiate color factor . the strength of the (strong) coupling for single-gluon exchange between two color charges is 21 × c1 × c2 × αs .1. in which a highmass lepton pair ℓ+ ℓ− emerges from q q¯ annihilation in a pp collision. it would have been more natural to absorb the factor 12 in a redefinition of the strong coupling αs and just let the product |c1 c2 | be known as the color factor). where c1 and c2 are the color coefficients associated with the vertices.Appendix F Spin. The simplest example to analyze is the Drell-Yan process. in QCD. in fact. the strength of the electromagnetic coupling between two quarks is given by: eq1 × eq2 × α. or − 13 ) and α is the fine structure constant. where eqi is the electric charge in units of e (that is eqi = + 32 .and Color-Averaged Cross Sections In QED.

4πe2q α2 tˆ2 + sˆ2 dσ .

.

(F. = CF .0.1) sˆ2 sˆ2 dtˆ .

25. Yan. The factors of 13 average over the initial q and q¯ colors. and the factor of 3 sum over q q¯ color combinations which 1 S. D. Lett. 25. 211 . M.qq¯→ℓ+ ℓ− where CF = 12 × 31 × 31 × 3 = 16 . Phys. Rev. 316 (1970) [Erratum-ibid. Drell and T. 902 (1970)].

the differential cross section for (massless) partonic subprocesses leading to jet pair production can be written.can annihilate to form a colorless virtual photon. The only difference between the two calculations is that we must average rather than sum over the color orientations of the quark and antiquark. CF = 21 × 13 × 81 × 8. and for the Compton process qg → qγ. the cross section for q q¯ → ℓ+ ℓ− . as . to lowest order in QCD. CF = 21 × 31 × 13 × 8. To LO QCD. In a similar fashion.5. is simply related to the cross section for e+ e− → q q¯ given in (3.82). Duplicating this reasoning we obtain for the annihilation process q q¯ → gγ.

παs2 ij→kl dσ .

.

= Σ .2) dt .0. (F.

9 s2 and for simplicity. − 9 u s t2 4 s2 + u 2 . Gluck. 9 t2 s2 27 ts 4 u2 + t2 . 9  t2  8 s2 4 s2 + u2 s2 + t2 + − . 1501 (1978).2 2 J. 212 . we drop carets for the parton subprocesses. F. E. 6 u t 8 s2   8 t2 + u2 u 32 t − + . D 18. Phys. 27 u t 3 s2 4  s u  s2 + u 2 + + . Owens. 2 s t u   2 3 t + u2 u 1 t − + .ij→kl s2 where gg→gg Σ = Σgg→qq¯ = Σqq¯→gg = Σgq→gq = Σqi qj →qiqj = Σqi qi →qi qi = Σqi q¯i →qi q¯i = Σqi q¯i →qj q¯j =   tu su st 9 3− 2 − 2 − 2 . 9 t2 u2 27 tu   8 u2 4 s2 + u2 u2 + t2 − + . Rev. Reya and M.

using (2.e.Appendix G Monojets /T ) with balancing transverse Events with a single jet plus missing energy (E momenta (so-called “monojets”) are incisive probes of new physics. couplings.4. the differential cross section follows from the relevant expression of conventional QED in Table 3. keeping only transverse Z’s).1. Ignoring the Z mass (i. For example. and mixings..94) and . supplemented by the appropiate color factor. In the standard model the dominant source of this topology is ij → kZ 0 followed by Z 0 → ν ν¯.

.

  dσ .

.

πα2 uˆ sˆ dσ .

.

1) = = 2 − − sˆ sˆ uˆ dtˆ .0. (G. .

γe− →γe− dtˆ .

γe− →γe− L L R R we obtain .

 2   dσ .

.

gs2 g 2 21 − 23 sin2 θw uˆ sˆ 1 = CF − − 16π cos θw sˆ2 sˆ uˆ dtˆ .

guL →ZuL     2 1 uˆ sˆ 1 gs2 g 2 21 − 23 sin2 θw . − − = 6 16π cos θw sˆ2 sˆ uˆ .

 2   1 uˆ sˆ dσ .

.

− − = cos θw sˆ2 sˆ uˆ dtˆ . 1 gs2 g 2 − 21 + 13 sin2 θw .

gdL →ZdL 6 16π .

dσ .

.

dtˆ .

gu R →ZuR  2   1 gs2 g 2 − 32 sin2 θw 1 uˆ sˆ = .2) (G.0.0.0.4) .3) (G. − − 6 16π cos θw sˆ2 sˆ uˆ 213 (G.

and .

dσ .

.

dtˆ .

2).0. and (G.5) Now.5) the contributions to gq → Zq become .0.gd R →ZdR 1 gs2 g 2 = 6 16π 1 3 sin2 θw cos θw 2   1 uˆ sˆ .0.0.4).3). (G. (G.0. combining (G. − − sˆ2 sˆ uˆ (G.

  dσ .

.

(G. π αs α 41 − 23 sin2 θw + 98 sin4 θw 1 uˆ sˆ = .6) − − 6 sˆ2 sˆ uˆ sin2 θw cos2 θw dtˆ .0.

gu→Zu .

  π αs α 41 − 13 sin2 θw + 92 sin4 θw 1 uˆ sˆ dσ .

.

7) − − 6 sˆ2 sˆ uˆ sin2 θw cos2 θw dtˆ .0. = . (G.

for q q¯ → Zg we obtain .gd→Zd Finally.

  tˆ dσ .

.

4π αs α 14 − 32 sin2 θw + 98 sin4 θw 1 uˆ + = 9 sˆ2 tˆ uˆ sin2 θw cos2 θw dtˆ .

8) and .u¯u→Zg (G.0.

dσ .

.

4π αs α = .

as obtained from numerical integration of X Z 0 dσ dY fi (xa . 4.14 we show the invariant mass distribution of the Z + jet final state. ˆ 9 dt dd→Zg ¯ 1 4   − 31 sin2 θw + 92 sin4 θw 1 uˆ tˆ + .9) In Fig. W ) = Wτ dW −Y max ijk . sˆ2 tˆ uˆ sin2 θw cos2 θw (G. W ) fj (xb .0.

Z ymax +Y 1 dσ .

.

× dy .

dtˆ ij→Zk cosh2 y −(ymax +Y ) Z Ymax dY fi (xa . W ) fj (xb . W ) + 0 # .

Z ymax −Y 1 dσ .

.

(G.0. .10) × dy dtˆ .

214 . S. B 821. for y1 . Lust. y2 < 1.06%. Nawata. Stieberger and T. Goldberg. Taylor.00 ± 0. H. Anchordoqui. S. A.ij→Zk cosh2 y −(ymax −Y ) /T is 20. Phys. D. Nucl.1 The branching fraction of Z into E 1 L. R. 181 (2009).

Recall that the outgoing ν¯e is described by v(k ′ ).0. According to the Feynman rules. H. (H. it must be drawn using only particle lines.1). The particle four momenta are defined in (H.1.4) .0.2) where the spinors are labeled by the particle momenta. and so the outgoing ν¯e is shown as an incoming νe .1) is the model reaction for weak decays.0. 5 ′ ′ (2π) 2E 2ω dQ = 215 (H.0. and the Feynman diagram is shown in Fig. 1 |M|2 dQ . (H. The muon decay rate can now be obtained using (A.0.0.8).Appendix H Muon Decay Muon decay µ− (p) → e− (p′ ) ν e (k ′ ) νµ (k). (H. The invariant amplitude for muon decay is M= G √F [u(k)γ µ (1 2 − γ 5 )u(p)] [u(p′ )γµ (1 − γ 5 )v(k ′ )] .3) dΓ = 2E where the invariant phase space is d3 k ′ d3 k d 3 p′ (2π)4 δ (4) (p − p′ − k − k ′ ) (2π)3 2E ′ (2π)3 2ω (2π)3 2ω ′  1 d 3 p′ d 3 k ′ = Θ(E − E ′ − ω ′)δ (p − p′ − k ′ )2 .

and so in (a) the outgoing ν¯e is shown as incoming νe . k 0 ≡ ω.5) δ(ω − |~k|) = 2ω 2|~k| Using (H. (H.1: Tree level diagram of muon decay. 1X |M|2 ≡ |M|2 2 spin 2 X  1 GF √ [u(k)γ µ (1 − γ 5 )u(p) u(p)γ ν (1 − γ 5 )u(k) = 2 2 spin X × [u(p′ )γµ (1 − γ 5 )v(k ′ )v(k ′ )γν (1 − γ 5 )u(p′) spin × = × −  2 GF √ Tr[6 kγ µ (1 − γ 5 )(p/ − mµ )γ ν (1 − γ 5 )] 2 ′ Tr[p/ γµ (1 − γ 5 ) 6 k ′ γν (1 − γ 5 )]  2 1 GF √ {Tr[6 kγ µ (1 − γ 5 )p/γ ν (1 − γ 5 )] 2 2 ′ Tr[p/ γµ (1 − γ 5 ) 6 k ′ γν (1 − γ 5 )] ′ mµ Tr[6 kγ µ (1 − γ 5 )γ ν (1 − γ 5 )]Tr[p/ γµ (1 − γ 5 ) 6 k ′ γν (1 − γ 5 )]} . 1 = 2 (H. with p0 ≡ E. and so on. According to the Feynman rules introduced in Chapter 3.0. the diagram must be drawn using only particle lines.0.0.2) and neglecting me we find the spin-averaged probability.6) 216 .e− νµ W νe e− νµ ≡ µ− ν¯e W µ− (a) (b) Figure H. and where in reaching the last line we have performed the d3 k integration using Z Z Z 4 2 d k Θ(ω)δ(k ) = dω d3 k Θ(ω) δ(ω 2 − |~k|2 ) Z Z Z 3 1 d k 3 = d k dω . In (b) we show the time direction of the antiparticle’s four-momentum.

10) 2E ′ ω ′ to perform the integration over the opening angle θ between the emitted e− and ν¯e and obtain dΓ = G2F dE ′ dω ′ mµ ω ′ (mµ − 2ω ′) .0. Tr[6 kγ µ (1 − γ 5 )γ ν (1 − γ 5 )] = Tr[6 kγ µ γ ν (1 + γ 5 )(1 − γ 5 )] = 0. 0). k ′ = mµ ω ′. Because we neglected the mass of the electron p − k ′ = p′ + k and k 2 = p′2 = 0. 0. p) = (m2µ − 2mµ ω ′ )mµ ω ′. therefore 2(k . 2π 3 (H. (H.0.Substituting (B. 4 (H.12) (H.13) These limits are easily understood in terms of the various limits in which three-body decay µ → e¯ νe νµ becomes effectively a two-body decay. 0 ≤ E ′ ≤ 21 mµ .11) The delta function integration introduces the following restrictions on the energies E ′ .0. p) = 64G2F (k . 0. ω ′ . p) .0. p) = (p′ + k)2 (k ′ . Gathering these results together. we have p . p′ )(k ′ . p) = (p − k ′ )2 (k ′ ..10) in the first term we obtain  2 GF |M|2 = 256 (k . p) = [p2 − 2(p . For 217 . stemming from the fact that −1 ≤ cos θ ≤ 1: 1 m 2 µ − E ′ ≤ ω ′ ≤ 12 mµ . (H.9) Now. we can replace d3 p′ d3 k ′ by 4πE ′2 dE ′ 2πω ′2dω ′ d cos θ. the decay rate in the muon rest frame is G2F d3 p′ d3 k ′ mµ ω ′(m2µ − 2mµ ω ′) 2mµ π 5 2E ′ 2ω ′  × δ m2µ − 2mµ E ′ − 2mµ ω ′ + 2E ′ ω ′(1 − cos θ) . p′ )(k ′ . so 2(k . where p = (mµ .0.7) because the trace of the second term vanishes. p′ )(k ′ . dΓ = (H. p) .8) In the muon rest frame. p′ )(k ′ .0.0.e.0. i. and use the fact that 1 δ(· · · + 2E ′ ω ′ cos θ) = δ(· · · − cos θ) (H. k ′ )](k ′ .

11) mµ G2F dΓ = dE ′ 2π 3 Z 1 m 2 µ 1 m −E ′ 2 µ dω ′ω ′ (mµ − 2ω ′ )   G2F 2 ′2 4E ′ = . (H.12) yields ω ′ = mµ /2.0.example when the electron energy E ′ vanishes. which is expected because then the two neutrinos share equally the muon’s rest energy. We find GF ≃ 10−5 /m2N .16) where we have chosen to quote the value with respect to the nucleon mass. To obtain the energy spectrum of the emitted electron.0.197019 ± 0.0. 3− m E 12π 3 µ mµ (H.0.15) Inserting the measured muon lifetime τ = (2. we can calculate the Fermi coupling. we calculate the muon decay rate 1 Γ≡ = τ Z mµ /2 dE ′ 0 G2F m5µ dΓ = . 218 . dE ′ 192π 3 (H.14) This prediction is in excellent agreement with the observed electron spectrum.0.000021) × 10−6 s. Finally. we perform the ′ ω integration of (H. (H.

219 .and right-handed longitudinally polar− ized electrons and τL. see e.2)  (I. T.5. Paul.0.g.0.R left.and right-handed τ ’s whose polarization can be experimentally analyzed by observing the decay τ → πντ .91) is valid near q 2 ≃ z with  Z 1 Z 0 ¯ dσ(e+ e− → f f¯) dσ(e+ e− → f f) − d cos θ AFB = d cos θ d cos θ d cos θ −1 0  −1 ¯ × σ(e+ e− → f f) .3) + − Aτ = σ e e → τL− τ +  + − −σ e e → τR− τ +   . In the above asymmetries θ is the angle between the produced fermion f and the incoming e− e− L.. CERN-THESIS-98-008.Appendix I Asymmetries at the Z-pole Equation (5. C.R represent left.0. + − + − = σ e eR → f f¯ − σ e eL → f f¯ σ e+ e− → f f¯ . σ(e+ e− → τ τ¯) .1) ALR and      .1 1 For details. (I. (I.

0236449 and PHY-0757598) and the US Department of Energy (DoE Grant DE-FG02-95ER40896). Mario Ciofani. and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF or DoE. Carlos Nu˜ nez. and Brian Vlcek) carefully read a draft of these notes and pointed out many misprints and unclear passages. and Tom Paul for entertaining discussions. Concha Gonzalez Garcia. Maddie White. Haim Goldberg. Leslie Wade. Daniel Gomez Dumm. The Fall’09 students (Sydney Chamberlin.Acknowledgments We would like to thank Carlos Garcia Canal. This work has been partially supported by the US National Science Foundation (NSF Grants OPP. Any opinions. 220 . findings. Russell Moore.