This action might not be possible to undo. Are you sure you want to continue?
An Introduction to the Standard Model of Electroweak Interactions
Giovanni Ridolfi1
CERN THDivision, CH1211 Geneva 23, Switzerland
IOn leave of absence from INFN, Sezione di Genova, Italy.
Contents
1 Introduction 2 Construction of the standard model 2.1 A gauge theory of weak interactions. 2.2 Masses .. 2.3 Summary Special topics 3.1 The scalar sector beyond the tree level 3.2 The SU(2) custodial symmetry 3.3 Axial anomaly cancellation. 3.4 Accidental symmetries Appendices 4.1 Renormalizability and power counting. 4.2 Nonunitarity of the Fermi theory . 4.3 Gauge theories . 4.4 The standard model lagrangian in renormalizable 4.5 Dimensional regularization 2 3 3 11 22 24 24
36
3
38 45 49 49
4
51
55 gauges .
58 62
1
Chapter 1 Introduction
The aim of these lectures is a description of the construction and the main phenomenological implications of the Glashow WeinbergSalam unified theory of weak and electromagnetic interactions (universally referred to as the standard model.) Basic knowledge in quantum field theory [1],[2] and elementary group theory [3] is assumed, as well as familiarity with the fundamental phenomenology of weak interactions [4]. No attempt will be made to give a full list of references. Such a list can be found in any standard text book of particle physics; see for example refs. [4][8].
2
1) From the experimental values of the muon and neutron lifetimes. see Appendix 4. Ve for the electron neutrino.2) (2. and it gives rise to a nonunitary S matrix (see Appendix 4. T(N) one obtains = 885.1).3) while the value a = 1. and so on. a gauge theory. However.1. The field theory defined by the interaction in eq.2).1. This lagrangian (often called the Fermi lagrangian) has the form of a sum of products of vector and axial vector currents. We will then require that the new theory reduce to eq. since it contains operators with mass dimension 6 (a necessary condition for perturbative renormalizability is that the lagrangian density contains operators with mass dimension less than or equal to 4.19703 ± 0.1.1.1.1) in the lowenergy 1Throughout these lectures.7 ± 0. 3 . For example.8 s T(jj) = 2.1 of the standard model A gauge theory of weak interactions Our starting point is the effective lagrangian that describes weak interaction processes at low energies.00004 s.09 (2.1) is manifestly not renormalizable.4) can be extracted from the measurement of hyperon decay rates. the terms responsible for nucleon f3 decay and for muon decay are! (2.239 ± 0. it contains all the physical information needed to build a renormalizable and unitary theory of weak interactions. (2. (2. particle fields will be denoted by the symbol usually adopted for the corresponding particle: e for the electron.1.Chapter 2 Construction 2.3). (2. in analogy with quantum electrodynamics (see Appendix 4. The idea is that of building a theory with local invariance under the action of some group of field transformations.
Both these steps can be performed with the help of the information contained in the Fermi lagrangian. which. In order to complete this program.11) will also be present.8) (2.tt[T+. The currents are in onetoone correspondence with the generators of the symmetry group. and are the corresponding generators.1. We observe that the current Jtt can be written as Ti1 (2.9) .1.T +_1( T1 + 'lT2)_[01] 0 0 "2 and Ti are the usual Pauli matrices. The hermitian conjugate current (2. Therefore.1. In the case of Jtt.7) where L= t (1 .limit.1. we must choose the group of local invariance. since gauge theories are known to be renormalizable. No other current must be introduced.12) We have thus interpreted the current Jtt as one of the three conserved currents of a theory with SU(2) gauge invariance.1. (2. in the sense that the local fourfermion interaction of the Fermi lagrangian will be interpreted as the interaction vertex that arises from the exchange of a massive vector boson with momentum much smaller than its mass.6) where 'l/Ji are the components of some multiplet of the (as yet unknown) gauge group. the Pauli matrices being the generators of SU(2) in the fundamental 4 . since (2.1.1. (2. They participate in the weak interaction via the current (2. In this way. and the mass of the intermediate vector boson will act as a cutoff that stops the growth of cross sections with energy. thus ensuring unitarity of the scattering matrix.5) We would like to rewrite Jtt in the form of a Noether current. and then assign particle fields to representations of this group. T]L = L.'5) ( ~) (~:) ' . both problems of renormalizability and unitarity will be solved. the current Jf = L. Let us first consider the electron and the electron neutrino. this can be done in the following way.10) will also participate in the interaction.ttT3L (2. in turn. form a closed set with respect to the commutation operation: the commutator of two generators is also a generator.1.
As we will see. For example. We will come back later to the problem of neutral currents. the experimental observation of phenomena induced by weak neutral currents is a crucial test of the validity of the standard model. so they must be assigned to the singlet (or scalar) representation. Of course. one for each of the three SU(2) generators.1. VeR and en. the electromagnetic current. the fields The chargedcurrent term Lc is usually expressed in terms of . The current Jf is a neutral current: it contains creation and annihilation operators of particles with the same charge (actually.1. For the moment.13) where we have introduced.representation. and second.1.1.14) where IfJ = rttDtt. This is for two reasons: first. and (2. and build a covariant derivative (2. where (2. of the same particle).1. as we will see) since it does not require the introduction of fermion fields other than the observed ones. The lagrangian L contains the usual kinetic term for massless fermions. and when it acts on the gauge singlet eR we have T. (2. we have Ti Td2. We must introduce vector meson fields Wi. this is not the only possible choice. The righthanded neutrino and electron components. that will end up with the inclusion of the electromagnetic current in the theory. Neutral currents do not appear in the Fermi lagrangian. the electromagnetic current involves both lefthanded and righthanded fermion fields with the same weight. the neutrino being chargeless. We are now ready to write the gaugeinvariant lagrangian for the fermion fields (which we assume massless for the time being): L iL IfJL + ilJeR IfJVeR + ieR IfJeR (2. and we have assigned the lefthanded neutrino and electron fields to an SU(2) doublet. o. The matrices Ti are generators of SU(2) in the representation of the multiplet the covariant derivative is acting on.17) to neutral current interactions.16) corresponds to chargedcurrent interactions. do not take part in the weakinteraction phenomena described by the Fermi lagrangian. no neutral current phenomenon is observed in lowenergy weak interactions.15) plus an interaction term L. we go on with the construction of our SU(2) gauge theory.18) . Wtt = J2(Wtt 5 ± 1 1 = zW tt)· 2 (2. a coupling constant g. + Ln. Notice also that the neutral current Jf cannot be identified with the only other neutral current we know of. as is customary in gauge theories. the electromagnetic current does not contain a neutrino term.1. but it is the simplest possibility (and also the correct one. when D" acts on the doublet L.
only the term Ln is modified.19) We have already observed that the neutral current Jf = L. g' is the coupling constant associated with the U(1) factor of the gauge group.igW. Since the SU(2) factor of the gauge group acts in a different way on lefthanded and righthanded fermions (it is a chiral group). in order to make Jf equal to the electromagnetic current.1. usually called the weak hypercharge. (VeLll1VeL .) eL eLII1 (VeLII1VeL + ~ EI1 [Y(L) This can be written as + eLll1eL) + Y(VeR)VeRII1VeR + Y(eR)eRll1eRj. and T3 = 0 for VeR and eR. which is the one that turns out to be correct. (2. The construction of the model can therefore proceed in two different directions: either we modify the multiplet structure of the theory. after the discovery of weak processes induced by neutral currents.We find (2. We will require our lagrangian to be invariant also under the U(1) gauge transformations (2. and Y('ljJ) is a quantum number.20) (2.22) zz 2' where Y is a diagonal matrix with the hypercharge values in its diagonal entries. and T3 = ±1/2 for VeL and ci.1. and we extend the gauge group in order to accommodate also the electromagnetic current in addition to Jf. We have now Ln gWI1 "2 3 Y being . it is natural to allow for the possibility of assigning different hypercharge quantum numbers to the left and right components of the same fermion field. . diagonal. it should be kept in mind that this was not at all obvious to physicists before the observation of weak neutralcurrent effects. We can now 6 .23) (2. The simplest way of extending the gauge group SU(2) to include a second neutral generator is to include an abelian factor U(1): SU(2) + SU(2) ® U(1).1.1./l'73L cannot be identified with the electromagnetic current.1. respectively. and the covariant derivative becomes Y DI1 = 811 . or we admit the possibility of the existence of weak neutral currents. to be specified for each field 'ljJ. Nevertheless. and correspondingly that the gauge vector boson Wf cannot be interpreted as the photon.24) where 'l1 is a column vector formed with all lefthanded and righthanded fermionic fields in the theory.1.I'y. We proceed to describe the second possibility.21) where 'ljJ is a generic field of the theory. A new gauge vector field EI1 must be introduced.ig'EI1 (2..
B!': B" = Att cos Ow . (2. (2.1.1. In other words. cos Ow = 0 g sin 0 w . This freedom can be used to fix arbitrarily the value of one of the three hypercharges Y(L).) Equation (2.1. The conventionally adopted choice is Y(L) With this choice.g cos 0 w =e . eq.31) (2. 2 2 9 sin Ow = g' cos Ow = e. 2 which is valid for any fermion. To do this. eq. (2.29) restricted = 1. with the photon field.1.34) T3+ =Q. Y(VeR) and Y(eR) so that Att couples to the electromagnetic current (2. (2.h. T3 9 sin Ow +~ g' cos Ow = e Q.assign the quantum numbers Y in such a way that the electromagnetic interaction term appear in eq. say Att.35) 7 .29) then reduces to Y (2.24) takes the form (2. provided we rescale g' by 1/ K. (2.29) The weak hypercharges Y appear in eq.28) where Q is the electromagnetic it must be charge matrix in units of the positron charge e.1. we must choose Y(L).1.27) In order to identify one of the two neutral vector fields.1.30) leptons is (2. Y(eR). but gauge invariance of the charged coupling requires Q1Q2 = 1.1. we find (2.25) (2. of eq.1.s. we have the freedom of rescaling the hypercharges by a common factor K.33) (For a generic doublet of fermions with charges Q1 and Q2 the r. In terms of the new vector fields Att. (2. (2.1. we first perform a rotation by an angle Ow in the space of the two neutral gauge fields Wf.1.33) becomes e(Q1Q2).1.1.1.23).26) (2.29) only through the combination Y g': thus. For example. ZV. which gives tg' 1.Z" sin Ow Wf = Att sin Ow + Z" cos Ow .1.1.32) to the doublet of lefthanded sin Ow  +tg 1. Y(VeR).1. (2.
(2. For example. down and strange quark fields respectively. We will do this in terms of quark fields.1.1.g. The second term in eq. dC/"SL. These terms induce processes at a rate which is not compatible with experimental observation. (2.39) and T+ = so that 0 cos ()c sin ()c 0 0 0 [ 000 1 . and it is an SU(2) singlet: it does not take part in electroweak interactions. S are the up.1. We must now include hadrons in the theory.1.42) The corresponding neutral current contains flavourchanging terms.1.This completes the assignments of weak hypercharge values to all fermion fields.1. such as e.38) where ()c is the Cabibbo angle (()c 13°) and U.1. Notice that the righthanded neutrino has zero charge and zero hypercharge.40) (2.37) The extension of the theory to more lepton doublets is straightforward.1. (2.44) 8 . taking as a starting point the hadronic current responsible for f3 decay and strange particle decays: (2.1.27) defines the weak neutral current coupled to the other neutral vector boson Zw It can be written as (2. Indeed. the ratio of the decay rates for the processes + 7[°e+v e K+ + 7[+ e+ «: K+ (2.36) where (2. d. We are tempted to proceed as in the case of leptons: define I"V (2.1.43) (2.41) This leads to a system of currents which is in contrast with experimental we find that observations. with a weight of the same order of magnitude of flavourconserving ones.
They suggested to introduce a fourth quark c (for charm) with charge 2/3 like the up quark.tL"2(1 '5)d sin ()c c. (2.'5)S. and therefore outside the energy range of available experimental devices.46) that is.tL"2 1 (1 '5)S (2. where now Q= and [~fl 1 .tL"2 (1 .1. The existence of the quark c was later confirmed by the discovery of the J/ 'ljJ particle.tL"2 (1 ./) CLSL .41). The current is usually written in the following form.sin ()c cos ()c o 0 0 0 0 0 (2.47) + cos ()c c. 1 The c quark being not observed at the time. (2.1.1.1. () ]2 c  1 cos? ()c ~ 1.is approximately r= [ SIn . (2. J. the chargedcurrent process (s + u) is enhanced by five orders of magnitude with respect to the neutralcurrent (s + d) one. The solution to this puzzle was found by S. The current Jtad can still be put in the form (2. Maiani..1.1.50) thanks to the fact that the upper right 2 x 2 block of T+ has been cleverly chosen to be an orthogonal matrix. Our theory should therefore be modified in order to avoid the introduction of flavourchanging neutral currents. tLT + L ( UL) d~ 9 + (.1. and to assume that its couplings to down and strange quarks are given by cos ()c u. analogous to the corresponding leptonic current: »: tL Jhad = (UL d') .1. Iliopoulos and L. they had to assume that its mass was much larger than those of u. (2. Glashow. d and s quarks.48) [0 0 T+ = 0 0 o o cos ()c sinO" . In fact.49) No flavourchanging neutral current is now present.1.51) .5)d 1 1 + sin ()c u. tLT + ( s~ CL ) .45) while observations give (2.
1.58) ..1.1.1.61) (2.62) 10 W:. d). (2.1. (2.attwVi . we must consider the pure YangMills term (2.54) Q f = ( ~~) .where (2.18).60) (2. wttwv j k: (2. ( ~t) . (c.1. To conclude the construction of the standard model lagrangian. Actually.1. whose origin will be investigated in section 3.52) The pairs (u.55) An equivalent (and often more useful) form of eq.1.24) is directly generalizable to include quark fields.1.56) The neutralcurrent lagrangian in eq.57) where BIW = att BV ..1. V becomes an n x n matrix.3. The final form of the chargedcurrent interaction term.1..53) where (2.av Btt tt Wi V . which we rewrite here: (2. and it must be unitary in order to ensure the absence of flavourchanging neutral currents..1.1.avwtt i + gEijk The corresponding expression in terms of the physical fields Ztt and Att can be easily worked out with the help of eqs.53) is (2. (2. is then (2. s) are called quark families.1. The structure outlined above can be extended to an arbitrary number of quark families. (2. With n families.25) and (2.59) (2.26). (2. there is a correspondence between quark and lepton families.1. including n families of leptons and quarks.1.
It follows that (2.W.2. Att) + ig cos Ow(W: z.1. In the Fermi theory.66) ~4 F ttyFttV . (mN 11 . wtzv __ g2 2 (2 .~ W:W. which is known to correctly describe lowenergy weak interactions.ZttV sin Ow.~Z ttyzttV . Let us consider the amplitude for downquark f3 decay. where FttV = attAV _ av Att zttV = attZv _ av ztt W~v = attwr . it is simply given by (2.] 2. .2) (we are neglecting Cabibbo mixing for simplicity).) (2.1.avw~.65) (2. the gauge vector bosons of weak interactions must have a nonzero mass.W.2. in order to make contact with the Fermi theory.1) In the context of the standard model.Wv+Ztt)] + h.~ W+ WttV 4 2 tty +ig sin Ow(W~ W~Av .We get W~v W~v W!v ~ ~ [W~ + ig sin Ow(W: Av .c. 1. The virtuality q2 of the exchanged vector boson is bounded from above by the square of the neutronproton mass difference. with amplitude (2.64) (2. [W~ + ig sin Ow(W: Av . We will also be able to set a lower bound to the mass of the W boson. q2 ::. wtAv +ig cos Ow (W~ W~Zv . Att) + ig cos Ow(W: Zv .63) Fttv sin Ow + ZttV cos Ow .Wv+Ztt)] + h. the same process is induced by the exchange of a W boson.(ApAu sin ' Ow + ZpZu cos" Ow + 2ApZu sin Owcos Ow) .67) + Fttvwtw~) + ZttVwtw~) (2gttVs" _ s" gVU_ u" gYP) [W:Wv.W.1.WpW.W.1.W.2 Masses Masses for the gauge bosons We will now show that.W. .ig(W:WvFttv cos Ow .c.
3 MeV)2. Let us investigate this point in more detail.2.6). (2. we also know that gauge theories are incompatible with mass terms for the vector bosons. unless some mechanism takes place to cut off this growth. the amplitude M will grow indefinitely with the energy E.2.5) So. (2. the polarization is given by EL = 2 (k /m. thus violating powercounting and making the theory unrenormalizable.2. E /m"() = k/m"( + O(m"(/ E 2 ). with a mass term for the gauge vector field: (2.7) The propagator /::1/LV has not the correct behaviour for large values of the momentum k: for k + 00 it becomes a constant. Clearly. (2. and rv GF_(_9 )2_1 m?v· V2 2V2 mw ~ 37.3) Recalling that 9 = e] sin Ow.3 GeV.mp)2 (1. is again unitarity of the scattering matrix. 0.(.2.3) gives us the lower bound (2.. 12 . (2.8) A massive vector (contrary to a massless one) may be polarized longitudinally.2.2. choosing the z axis along the direction of the 3momentum of the vector boson. In this case.9) where we have imposed the transversity condition k· E = 0 and the normalization condition E2 = 1. we know since the beginning that if weak interactions are to be mediated by vector bosons. and unitarity of the scattering matrix will eventually be violated. these must be very heavy. The amplitude for a generic physical process which involves the emission or the absorption of a vector boson with fourmomentum k and polarization vector E (k) has the form (2. mw must be non zero. Consider for simplicity the lagrangian of a pure abelian gauge theory.6) and work out the propagator /::1/LV for A/L in momentum space. For eq. I~I (2. On the other hand. m"( < 2.2) to be equal to the Fermi amplitude in the q2 + 0 limit.10 16 eV. such as the one defined by eq. We get (2. A related problem of a massive vector boson theory. 0. (2.2. if compared with the present upper bound on the photon mass. One possibility is to break gauge invariance explicitly and insert a mass term for the W boson by hand. but this leads to a nonrenormalizable theory. rather than vanishing as k:".2. eq.2.2.4) quite a large value.
and V (cp) is the socalled scalar potential.14) 1 1 +AvH(H2 A + G2) + 4(H2 + G2)2. If. that is. In principle. that change the phase of the complex field cp without affecting its modulus.2.12) All these minimum configurations (in the language of quantum theory. The lagrangian is given by c = _~FttV r. + (Dttcp)t Dttcp . coupled to one complex scalar field cp with charge e. they must be constant configurations.V(cp).11) where D" = att . because at the quantum level this is essentially what guarantees the renormalizability of the theory. on the other hand. we will come back to this point later. The simple theory we consider is scalar electrodynamics. 1 (2. the potential has now an infinite number of degenerate minima. Because of the requirement of translational invariance. for example. but the symmetry is not actually broken.To see how one can introduce a mass term for gauge vector bosons without spoiling renormalizability and unitarity. Now. The system will choose one of the infinite possible minimum configurations.2. This phenomenon is usually called spontaneous breaking of the gauge symmetry. In fact. and then we generalize our considerations to the standard model. the scalar potential takes the form V(cp) = (m2v + Av3)H + _(m2 + 3AV2)H2 + _(m2 + AV2)G2 2 2 (2.. invariance and renormalizability to be of the form We look for field configurations that minimize the energy of the system. but of course any other choice would be equivalent. so we can neglect the derivative terms and look for the minimum of the potential V. In fact.2A = 2 m2 _ 2V 1 2 . and all the properties connected with this invariance (such as. the field G could have been removed from the lagrangian by an appropriate gauge transformation. furthermore. and then shift it according to cp = (v+H)/V'i. 13 . if m2 ~ 0.2.2.13) where v is defined in eq. We introduce real scalar fields H(x) and G(x) by cp(x) = V'i [v + H(x) + iG(x)] . m2 < 0. which would instead be lost in the case of an explicit breaking of the gauge symmetry. Up to an irrelevant constant. given by all those field configurations for which I cp I = .2.12). we first consider a simple example where this happens. which is constrained (2.2. a gauge theory based on U(1) invariance. (2. we keep both Hand G in the lagrangian. then V has a minimum for cp = o. (2.10) by gauge (2. all these ground states) are connected by gauge transformations. It is important to stress this point. then m2 can no longer be interpreted as a mass squared for the field cp. we could have first applied a local gauge transformation to cp in order to make it real.ieAtt. For the moment. current conservation) are still there. the Lagrangian is still gauge invariant. we choose the one for which cp is real at the minimum. Let us now expand the field cp around one of the infinite minimum configurations.
where m'Y = ev and m't = 2AV2. It is well known that. (2. (2. and has therefore the correct sign to be interpreted as a mass term (remember that m2 is negative).8tt AV8 A ) tt v v tt + ~m2 Att A .20) . a gaugefixing term must be added to the lagrangian (obviously.2.2. precisely the result we were looking for.19) terms.21~(8ttAtt + ev~G?. A convenient choice for the ga uge. (2.2.2. + G2 + 2vH)Att + G2) . this was not necessary in the case of explicit gauge symmetry breaking). The propagators can be worked out from the quadratic collected in the first two rows of eq. in order to quantize a gauge theory.16) corresponds (2. contrary to what happened when we tried to break the symmetry explicitly.~~m2G2 2 tt 2 'Y Att .Using eq. After the reparametrization eq.~)kttkVl k2 .eAtt(H8ttG . We must now check that the appearance of a mass term for Att via the spontaneous symmetry breaking mechanism has not spoiled the renormalizability of our theory.16) contains a term ev8tt AG.~m2 H2 2 tt 2H +2e2(H2 AvH(H2 1 + ~8ttG8 G . because it mixes the gauge vector field Att with the unphysical field G. which after partial integration cancels the unwanted term in eq. the ID¢12 term takes the following form: (Dtt¢)t DII¢ r 1 8tt H8 H 2 tt eAtt(H8ttG + 1 1 8ttG8 G + _e2(H2 2 tt 2 + G2 + 2vH)Att + 1 _e2v2 Att Aw 2 A tt (2.evAtt8ttG We see that the gauge field Att has acquired a mass m'Y = ev.15). L = ~(8tt 2 AV8 A .16) has been carefully chosen in order to cancel the term proportional to Att8ttG in eq.2.17) where ~ is an arbitrary gaugefixing condition constant (the gauge parameter). Equation (2. AV2 = m2.2. we see that the terms proportional to Hand G2 vanish.12). (2. The gaugefixing lagrangian (2.2. which means that the field G is massless.15) .19). we will see in a moment how to get rid of it.13).2.~(H2 + G2?.2. eq.18) 2'" 2'" 'Y ' which gives a squared mass to the unphysical field G. Collecting the various terms.fixing term is LGF = . (2. Indeed. The term evAtt8ttG is unpleasant. (2. the lagrangian is given by: ~m._!_(8tt A )2 2 'Y tt 2~ tt +~8tt H8 H . Observe also that the gaugefixing lagrangian introduces a term 1 __ 1f:e2v2G2 = __f:m2G2 (2.G8tt H) .2.2.2.2.~m~ (2.16) to the (2.15).2. The coefficient of the H2 term is now (2m2)/2.G8tt H) (2. We get [g 14 ttV + (1 .
and has therefore three polarization states instead of two. When we let ~ tend to infinity. ttV _ lim ~~ (k) . (2.2. since the photon is now massive. (2. This singularity is located at the mass squared of the unphysical scalar field G. in the absence of gauge invariance and of the gaugefixing 15 . since we started with a complex scalar field and we end up with one real scalar.7): .21 ) for the two scalar propagators. scattering at tree level.24) 'Y and the Landau gauge. In fact. L k2m~ ( g ttV +~ kttkV) . However. Actually. It is easy to check this cancellation in specific cases. + H. ttV (2. in addition to the pole at k2 = an unphysical singularity at k2 = has now appeared.25) One last observation about the field G (x). The limit ~ + 00 is called the unitary gauge. ~ = 0.2. (2. The advantage of the unitary gauge is that the theory contains only physical degrees of freedom. since the propagator does not obey the powercounting rule. In order to perform this kind of checks. ~ = 1.k2 'l..20) in the form m. (2.20) takes the form of eq.2. the number of degrees of freedom stays the same. One can prove that the contributions of this term of the photon propagator to physical quantities are exactly cancelled by the contributions of G exchange. This terminology reflects the fact that.2. but in a hidden way: renormalizability must arise as a consequence of cancellations among different contributions to the same Green function. 2 m'Y ( ~too g ttV + 2m'Y kttkV) .2.23) The theory is still renormalizable.g. it corresponds to the gauge choice that eliminates G from the theory since the very beginning. the photon propagator eq. when ~ + 00 the gaugefixing condition reduces to G(x) = 0 (see eq. for which ~ttV _ 'l.2. (2. H. Observe that the photon propagator has now the correct behaviour 1/k2 at large momenta.2. and (2.2. it is useful to rewrite the propagator in eq.22) where the G propagator appears explicitly. Two common gauge choices are the Feynman gauge. (2.16)).2. ~m. such as e. The drawback is that in the unitary gauge renormalizability is not manifest at each intermediate step of a calculation. The field G(x) is called a wouldbe Goldstone boson. It looks like we lost a degree of freedom. which gives ~ttV F  k2 _ m2 'l_g__ .for the photon propagator. (2.
we want the photon to stay massless.29) is left unchanged by electromagnetic gauge transformations.1. with a few modifications that we now describe in detail. the fourth one being that corresponding to electric charge.2. This mechanism is known as the Higgs mechanism. CP2. Q2 are the electric charges of possi bili ties: CPl. (2.term.31 ) There are two where Ql. we must introduce scalar fields in the game.2.2. We have learned that.2. zeromass state. V. 16 . scalar potential consistent with gauge invariance and renormalizability is The most general (2.26) The Higgs mechanism takes place in analogy with scalar electrodynamics. which is always present when spontaneous symmetry breaking occurs.28) The value of the hypercharge of the scalar doublet cp is fixed by the requirement that the minimum configuration (2. and I v21 = we have used eq. Secondly.2.2. U(1)em.m = 2A 2 v . or. we must be careful not to break the U(1) invariance corresponding to electrodynamics. in order to break spontaneously a gauge symmetry. the scalar field must transform nontrivially under that part of the gauge group that we want to undergo spontaneous breaking. How should we do this in the standard model? First. G would have simply been a physical. 1) 2) VI = 0. The simplest way to do this is to assign the scalar field cp to a doublet representation of SU(2): (2.33) IVll=v.2. This corresponds to the requirement that correspond to the subgroup (2.30) or equivalently 0 (Ql o Q2 )(Vl) V2  (1/2+Y/2 0 1/2 0 + Y/2 )(Vl) V2  (0)0 ' (2.27) which has a minimum at 2 _ I cp 1= . 2 1 2 (2.2. in other words. V2 Y = +1 Y=1. This means that spontaneous symmetry breaking must take place in three of the four "directions" of the SU(2) x U(1) gauge group. = 0.34).32) (2. (2. It is possible to extend it to the standard model.
35) can be worked out the Higgs scalar H has a squared mass m~ = 2AV2. The standard model lagrangian in a generic renormalizable gauge is given in Appendix 4.1. there is always a linear combination of B" and Wf that does not receive a mass term. Note that the photon stays massless. but only if Y(¢) = 1 (or 1) does this linear combination coincide with the one in eq.38) (2.~ (H +v) ~ ( ~ ( :JH ) .1. if we work in the unitary gauge.36) We have where in the last step we have used eqs. the wouldbe Goldstone bosons. With the scalar field ¢ transforming as a doublet of SU(2).37) We see that the Wand Z bosons have acquired masses m2 = _g2v2 w 1 4 (2.25).2. since it also involves kinetic and interaction terms for nonphysical Higgs scalars.2. because it is nonlinear and contains all powers of the fields (Ii.2. we will use the unitary gauge (Ii = o. It is convenient.1. therefore (2.9 i i .34) with (li(x) and H(x) real.~ (H + v) ( ~~~f~i. The scalar potential takes the form V = (2AV2)H2 2 1 + AvH3 + AH4.We will adpot the first choice.2.4. (2.'l"2Btt 1 V2 ( H(x) + v 0 ) ~ (a~H ) . The term (Dtt¢)tDtt¢ using eq. This parametrization is not suited for renormalizable gauges. The lagrangian in a generic renormalizable gauge is much more complicated. It is described in Appendix 4. 17 .2. however.1.2. with Y = +1 and therefore Q1 = 1. (2.18).39) m.33).2. We will further assume that V2 is real and positive.£~ ) P ).4. = ~(g2 + g'2)V2. it is apparent that the fields (Ii can be transformed away by an SU(2) gauge transformation. (2. JC/':::)/2Z (2. We get (a tt  . We can reparameterize ¢ in the following way: (2. in fact.g' ) 'l"2T Wtt . 4 ' 1 (2.34) with (Ii = O. (2.26) and (2. (2. In this section.25).1. Q2 = O.
45) (2.2. Let us call u.2.41 ) and cannot be invariant under a chiral transformation.2.2.3) and (2. a transformation that acts differently on lefthanded and righthanded fields. in principle. and (2. it must) be included in the lagrangian. so does ¢>c exercise. the vacuum expectation value of the neutral component of the Higgs doublet. gaugeinvariant'' and renormalizable.43) £~adr = _(Q' ¢ h'n D' + tJ' ¢t h'b Q') .46) tensor.adr is Lorentzinvariant. (2.42) term can be added to the lagrangian: (2.J2 (2.2. In fact. We first consider the hadronic sector. the (2. = E¢>*. there is no reason why only downtype quarks should be rotated.2.The value of v. check it as an D R' 2If ¢> transforms as an SU(2) doublet.2.38).2. The matrices h~ and h'n can be diagonalized by means of biunitary transformations: u h U = Vuth' U yR L h .f R· (2. Masses for hadrons and flavourmixing Fermion masses are also forbidden by the gauge symmetry of the standard mass term for a fermion field 'ljJ has the form model. YG. and using the measured valued of the Fermi constant.44) It easy to check that £'y. We have seen in section 2.(Q' ¢c h~ U' + ()' ¢! h'~ Q').40) The value of the Higgs quartic coupling A (or equivalently the Higgs mass) is not fixed by our present knowledge. that is.22GeV.2. and therefore it can (actually. We also define Q'f = ( A Yukawa interaction ~1 ) D'i .1 that the interaction lagrangian is not diagonal in terms of quark fields with definite flavours.d. this difficulty can be circumvented by means of the Higgs doublet ¢.f and d' f the fields that bring the interaction terms diagonal (the index f runs over the n fermion generations). The gauge transformations of the standard model are precisely of this kind. where h~ and h'n are generic n x n constant complex matrices in the generation space. Again. where E is the antisymmetric 18 . We get V= ~~246.yDth' yD D= L (2. can be obtained combining eqs.
51 ) where (2..47. A generic n x n unitary matrix is formed with n2 independent real parameters. chosen so that are diagonal with real.c y hadr = __'2 (v 1 V L.D. (2. The only term in the lagrangian which is affected by eqs.48) are obviously global symmetry transformations of the free quark lagrangian.. nonnegative Now.2.where VLURD are unitary matrices. their number is 1 np = n 2 .2. as we already discussed in section 2 in the case of two fermion families. and there are as many as the coordinate planes in N dimensions: (2.2.48) is the chargedcurrent interaction. To conclude this subsection.54) 19 . (2. we define new quark fields u and d by entries. We can now identify the quark masses with f _ vhb (2. eq.2. Indeed. because of the universality of the couplings of fermions of different families to the photon and the Z.2.52) The matrix V is usually called the CabibboKobayashiMaskawa (CKM) matrix. (2. The matrix Venters the standard model lagrangian as a fundamental parameter.2.2.2.2.47. we now determine the number of independent parameters in the CKM matrix.50) mu . (2. They also leave unchanged the neutralcurrent interaction term. eqs.2.JD + hfU fi/ uf) .49) where hb.43) becomes .48) In the unitary gauge.2. + H) f=l ~ (hf (if df L.D are the diagonal entries of the matrices hU. on the same step as masses and gauge couplings.47) (2. (2.V2' Since the matrices VLURD are constant in spacetime.2.53) The remaining parameters are just complex phases. The values of its entries must be determined from experiments.2. (2.2. Some (nA) of them can be thought of as rotation angles in the ndimensional space of generations. It is a unitary matrix. we find (2. because the up and down components of the same lefthanded doublet are transformed in a different way.nA = "2n(n A + ) 1.2. and its unitarity guarantees the suppression of flavour changing neutral currents.
2. with one or two fermion families. (2.60) This puts the Yukawa interaction in diagonal form. £~Pt = _ L h~ (Lf f=l n ¢e~ + e~¢t Li). E· £~Pt = (L'¢h~E' + E'¢th'kL').2.56) Masses for leptons The same procedure can be applied to the leptonic sector. There is however an important difference. leaves the charged interaction L' .2. since JiePt = in the quark sector. the CKM matrix has four independent parameters: three rotation angles and one complex phase. n f=l 20 . the CKM matrix can be made real.2. which corresponds to np = 1.2. The number of really independent complex phases in V is therefore np = np .55) Observe that.jand T). (2.59) (2. (2.2. the total number of independent parametersi in the CKM matrix is (2.tt T+ L= L v{ .(2n  1) = ~(n 2 .58) The difference with respect to the case of quarks is that now we have the freedom of redefining the lefthanded neutrino fields using the same matrix VLE that rotates charged leptons: (2. and the 1 accounts for the fact that the entry corresponding to the intersection of the row and the column cannot be rotated twice. The first case with nontrivial phases is n = 3. Everything is formally unchanged: upquarks are replaced by neutrinos and downquarks are replaced by charged leptons (e. however. j. In the general case.1 phases are eliminable: in fact.tt e{.2.61 ) term un(2. In the standard model with three fermion families.62) but. which leads to considerable simplifications: as we have seen. there is no Yukawa coupling involving the conjugate scalar field ¢c.tt T+ L' = L . h' . Therefore.2. that can be rotated to eliminate the phase of one row and one column of V. there are n uptype quarks and n downtype quarks.Some of the np complex phases. This means that 2n . and there is only one matrix ofYukawa couplings. can be eliminated by redefining the lefthanded quark fields. righthanded neutrinos have no interactions.57) which can be diagonalized by means of a biunitary transformation (2. contrary to what happens changed.1)(n  2).
because the Yukawa coupling matrix can be diagonalized by a global transformation under which the full lagrangian is invariant. not only the overall leptonic number. As a consequence. using eq. the details are given in Appendix 4.64) As in the case of vector bosons. in the leptonic sector there is no mixing among different generations. we find (2.63) thus allowing the identifications (2.34). in renormalizable gauges there are also interaction terms between quarks and nonphysical scalars. In fact. 21 .2. (2.2.2. This is due to the absence of righthanded neutrinos. The values of the Yukawa couplings h~ are determined by the values of the observed lepton masses.4. but also individualleptonic flavors are conserved.In other words.
] (2. WtAv + Fttvwtw~) +ig cos ()w (W~W~ZV .~sirr' ()w .3 Summary model lagrangian in the unitary gauge is given by To summarize.(ApAu sin ' ()w where + ZpZu cos" ()w + 2ApZu sin ()w cos ()w)  t W:W.~ w+ WttV 4 4 2 tty +ig sin ()w (W~ W~Av .3 .WpW.~ F ttyFttV .3.W.1) where • Lkin is the free fermion lagrangian: n Lkin = L f=l [vf i~ vf + (if (i~ .mb) uf + (if (i~ .3.3.5) is the pure YangMills lagrangian: . (2.3) • L.5) + (if 1 + ~ sin ' + 15) df 1 z». () cos w sin w f=l t [vf . coupling: • Lem is the electromagnetic (2.wtzv g2 __ (2gttVs" _ s" gVU_ u" gYP) 2 (2 .3.2) The index f labels the n fermion families.5) uf vf + (if Itt ( Itt (1 + l sirr' ()w ()w + 15) ef + • LYM fil Itt (1. Neutrinos are assumed massless. the standard (2.3.w.6) + ZttVwtw~) [W:Wv.~ Z ttyzttV . (2.4) • Ln is the neutralcurrent Ln interaction term: Itt (1 = 4 ()e .m£) df] .7) 22 .m~) ef + fi/ (i~ .2.3. is the chargedcurrent interaction term: (2.
g. the Fermi constant GF and the ZO mass mz. m~.10) g' tan Ow = 9 (2.3.3.3. 4 (2. g.9) The parameters appearing in LSM are not all independent.3.12) The free parameters in the fermionic sector are the 3n masses and the (n . V are often eliminated in favour of the electromagnetic coupling CYem.• The Higgs sector provides a term 1+(H) v • The Yukawa coupling LYukawa 2 1 m 2 2 H 2 3 1 4 H }. mb. 23 . We have G __ F V2v2' I_ (2. However.. g'. since The gaugeHiggs sector is entirely (2.vH }.11) and 9 sin Ow = g' cos Ow = e. This gives a total of 17 free parameters for the standard model with three fermion generations. m£. which are measured with high accuracy.8) is given by (2. specified by the four parameters I g.H.. v.1)2 independent parameters in the CabibboKobayashiMaskawa matrix V.3.
which we briefly recall. Functional derivatives of Z[J] with respect to J at J = 0 give the Green's functions of the theory. One introduces the functional (3. beyond the classical level. for this reason.Chapter 3 Special topics 3. ¢c as One then defines the classical field (3.4) The effective action has an expansion in powers of the classical field.1. so we do not introduce here sources for the other fields in the theory).1.2) is the generating functional for connected Green's functions. It can be shown that the functional W[J] = ilogZ[J] (3.5) 24 .1.1.1 The scalar sector beyond the tree level Effective action and effective potential In this section we will study the scalar sector of the standard model.1.1) where J(x) is a classical source with the appropriate gauge transformation properties (we are only interested in the scalar sector.3) ¢ ( ) = JW[J] cx JJ(x) and the effective action I'[ ¢c] as = (Ol¢(x) 10)J (OIO)J (3. This is most conveniently done in the context of the generating functional formalism. and in particular the phenomenon of spontaneous breaking of the gauge symmetry. Z[ J] is called the generating functional. (3.
1.... one has (3.1..1. xn): (3. (3. contain instead two or more derivatives of <Pc.1. as can be read off eq. for J = 0.1. The functional r[<Pcl is the appropriate tool to study spontaneous symmetry breaking.1. the condition for spontaneous symmetry breaking is that <Pc is different from zero even when the source J is set equal to zero. (3.9) The first term in this expansion is usually written as (3.6) We conclude that spontaneous symmetry breaking takes place when the classical field that minimizes the effective action is different from zero.6) reduces to (3. Consider now the Fourier transforms of the functions r n(Xl. The minimum condition eq.12) if we require translational invariance of the vacuum state. 25 . xn) can be shown to be the connected.whose coefficients rn(Xl' .7) and expand rn in powers of momenta around Pi = 0.1.1.3). oneparticle irreducible Green's functions of the theory. (3.8) The effective action becomes (3.10) where (3.1. since it does not contain derivatives of the classical field. originating from higher powers of momenta in the expansion of r n. The following terms.11) is called the effective potential of the theory. . In fact. On the other hand.
Consider for example a theory with a single real scalar field cp. This is no surprise: these terms are proportional to cP~ and cP~ respectively. We get 1Tfinite VI _  (3.1)(n .327r2 4 ~ 00 00 zfinite VI _  (1)nzn n(n .1. we find v.18) It is now easy to sum the series by shifting the summation index to n + 1 and n + 2 in the second and third term.16) One sees immediately that the terms corresponding to n = 1 and n = 2 are divergent. (3. 2) n 1 f dk 1 (27r)4 (k2m2+iTJr· 4 (3.13) r _ 2n(O) = i s. using the properties of the 1 r function and defining z = 3Acp~/m2.2) m42n r(n) .5.2). and a treelevel potential given by Vo(cp) = ~m2cp2 + ~Acp4.17) or. by taking the sum of all diagrams with an arbitrary number of external scalar lines and zero external momenta. and by adding and subtracting the missing n = 1.1. and by 2n because there are 2n identical vertices in the diagram.1. The oneloop correction to the scalar potential is therefore given by V1(CPc)="2. i 00 ( 3ACPc .1 n.1. 2 4 The oneloop Green's functions at zero external momenta are given by (3. (iA 4! 4 ) n f dk ( i (27r)4 k2 _ m2 4 + iTJ )n (3.14) while Green's functions with an odd number of external lines are obviously zero.finite 1 = i_i_ 2 (47r)2 n=3 f= (3Acp2)n (1)n c n r(n . (4.1. The combinatorial factor Sn is (3. (3. m4 .1. Let us first take care of the finite part..2 1] .2) n m z=(1)nzn 2 647r n=3 [1 2 +n. this number must be divided by 2n because there are two external lines for each vertex.2 terms.Effective potential for a real scalar field The effective potential can be computed directly. and the divergences must undergo the usual procedure of mass and coupling constant renormalization.15) and can be determined in the following way: there are (2n)! ways of assigning the external momenta to the vertices.1. The loop integrals can be performed using eq.19) 26 .
we have to compute explicitly the loop integrals in eq.2 + B 'i'e' '"4 (3. and since the finite part of the oneloop corrections starts with ¢~.1. (3. we could require that (3.1. In general.¢~) 27 (~ . we find V1 div = .26) . E 1 (3.4 'i'e' (3.20) is quadratically divergent: if we were to regularize the integrals by simply imposing an ultraviolet cutoff A on the modulus of the loop momentum k. A modified version of this (MS) consists in subtracting the term proportional to .1  1 Tfinite vI . perform the socalled minimal subtraction (MS).5. (3. both are divergent in the physical limit. (3. A + 00 for the cutoff regularization. or d + 4 in dimensional regularization.1.4 is subtracted. This fact is characteristic of scalar mass parameters.div 12 i [(3). after regularization.. this prescription simply means that the counterterms must be exactly equal and opposite to the divergent part.64 7r2 [6). + log( 47r) + log ~: ) 1' (3. For example.. We notice that the first term in eq. the divergent part of the oneloop potential takes the form v. (4.1.23) (3..22) hold for the treelevel potential.1.24) so that in this case Another possibility is to amounts to computing the counterterms in such a way renormalization prescription v.div 1 = A 'i'e ". namely v. Then. and then fixing the that only the pole in d . different choices correspond to different renormalization schemes. the renormalizability of the theory manifests itself in the fact that the only divergent diagrams correspond to terms which are already present in the bare lagrangian.¢~m2 1 + 6)'¢~ (m 2 + ~).1. we would find a term proportional to )'A2¢~.25) where the spacetime dimension is d = 4 . one must add suitable counterterms in order to cancel the divergences. e. The finite parts of the counterterms are arbitrary. which corresponds to a quadratically divergent radiative correction to the mass of the scalar field ¢e.2).22) Since eqs.1. In this case.20) The renormalization procedure requires that a regularization prescription is given in order to give mathematical meaning to the divergent integrals.2 _ B". Using again eq. This prescription divergent part in dimensional regularization.ct 1 = A 'i'e ".1. and consequently to different definitions of the renormalized parameters.2E.¢2) e f (27r)4k2m2+iTJ+2 dk 4 1 ~ (3).1.21) where A and B are functions of ).g. We must give some renormalization prescription to fix the finite counterterms. + log(47r). m and of some parameter which defines the regularization prescription.¢2) 2 e f (27r)4(k2m2+iTJ)2· dk 4 1 1 (3.20).Let us now consider the divergent part: v.1.
32) tells us that the effective potential of the original theory can be obtained by computing the oneparticle (or tadpole) amplitude of the shifted theory and integrating it with respect to the shift.28) and we have dropped constant terms. We now turn to the oneloop term.j2 2 ~l 2' (3.where j.o) r: _ = .MS 1 = _1_ ( 647r2 m 2 + 3A!. the effective potential receives contributions also from fermion and vector loops.1. 1 1 (3.33) (3. There is only one diagram to be computed. with one external 28 .30) where the Green's functions f'~ can  be computed in terms of f' no 1 From eq.j is an arbitrary mass parameter which must be introduced in dimensional regularization in order to keep the coupling constant A dimensionless. These contributions can be computed in the same way as the scalar one.I + log( 47r).27) where we have used the identity (3. Now. The treelevel potential of the shifted theory is V~(cp) The treelevel tadpole is therefore = "2m2(cp + w? + 4A(cp + W)4.n! 00 1_ rn(O)cp~ = V(CPc). obtained from the original one by shifting the scalar field by an arbitrary quantity w: cp+cp+w.1. we find v.31) r~ (w. integrated in w between 0 and CPc gives minus the treelevel potential (3. and therefore 0) = n=l 1L . The corresponding effective potential is (3. Fortunately. which allows one to obtain all contributions to the oneloop scalar potential in a very simple way.13) as expected. Let us see explicitly how this works. In more complicated theories.1.34) which.r n(O) nwn00 n.1.2)2 'f'c [1og m + 3ACP~j.30) we find (3. (3. but the calculations are quite tedious and complicated.1.29) (3.1.1. there is a much cleverer technique.32) Equation (3.1. like the standard model.1. 10 dwr~(w. Adding all together.1. Consider a new theory. (3.1. we simply subtract the term proportional to 1/ E .
38) In the standard model. which we write in terms of four real scalar fields ¢i: (3. (3. the effective potential receives contributions from the scalar sector.~):¢l V(¢. In the shifted theory.1.35) r~(w. It can be shown that the gauge dependence of the effective potential is governed by the equation [~:~ +C(¢. and therefore ddk i (3.1. 0) = 3AW (27r)d k2 _ m2 _ 3AW2· f Using the results of appendix 4. Equation (3.1.1.~) is a function which can be computed order by order in perturbation theory. the vector boson sector.j is the renormalization the MS subtraction.1. eq.1. The effective potential is a gaugedependent quantity. After performing (3.39) (we drop the suffix c on from now on). The effective potential in the standard model The procedure outlined at the end of the previous subsection can be applied to the standard model.37) which is the same result obtained with the direct calculation.5 we readily find r~ 0) (w. in particular. we find scale introduced by dimensional regularization. (3.40) where ~ is the gauge parameter and C (¢. the FaddeevPopov ghost sector and the fermion sector: V1(¢) = Vs(¢) + Vv(¢) + Vg(¢) + VF(¢) (3. = where j.line and one internal propagator.~) = 0. tells us that V is gaugeindependent at 29 . the mass of the ¢c field is m2 + 3AW2.27). The scalar field is now a complex doublet. and the ¢~ vertex is 3AW (the factor 3 is due to the fact that the three lines are identical).1.40).
(3. A simple way to circumvent this difficulty is to choose Wi = 0 for all i except one of them. The only term of the lagrangian we need is the scalarscalarvectorvector term that appears in the squared covariant derivative of the Higgs doublet. After the shift ¢i + ¢i + Wi. ¢1. In the m2 < 0 case. with mass m2 +3AV2. the treelevel potential (3.1. with mass m2 + AW2. the vacuum expectation values of the fields ¢i are all zero. and eq. due to the fact that the mass term in eq. this term contains both the mass terms for 30 . and a contribution from the three wouldbe Goldstone bosons. say W3 = W (the reason of this choice will become clear later.43) is easily intrerpreted: there is a contribution coming from the physical Higgs boson. this result holds in both the m2 > 0 and m2 < 0 cases.42) describes three real scalars. except that all four contributions must be taken into account. We will compute V(¢) in the Landau gauge ~ = 0.42) is not diagonal. (3. We now turn to the contribution of vector bosons. the minimum of the treelevel potential lies at ¢2 = v2. In the first case. after shifting the fields ¢i. In both cases. since now eq. it does not affect the final result). and one real scalar. The oneloop contribution is obtained in the same way as in the case of the real scalar field. there is no spontaneous breaking of the gauge symmetry. More specifically. ¢2 and ¢4. we stress the fact that the result in eq.43) where ¢2 = ¢i¢i. In fact. that is by computation of the oneloop tadpole diagram. The result is therefore Vs(¢) = (3.1. in this case.43) (as well as all the other contributions.1. This choice simplifies considerably the calculation.1. (3. the oneloop effective potential has the same form. (3.1. of course. ¢3.1. the ghost contribution Vg (¢) vanishes. whose masses vanish at the minimum of the treelevel potential. Secondly. The calculation is now exactly analogous to that of a single scalar field.its minimum. we observe that the same result could have been obtained without any specific assumption about the shift variables Wi. with mass m2 + 3AW2. to be computed below) is independent of the values of m2 and A. and the scalar masses are all equal to m2. Vv (¢). Some comments are in order. where 8V/8¢ = o. The trilinear couplings ¢3¢j¢j are simply AW for i =1= 3 and 3AW for j = 3. It is immediate to check that integrating the treelevel tadpole with respect to Wi and summing over the index i gives back the treelevel potential. First of all. We begin by computing the scalar contribution. A complication arises here.1.42) + ~A(¢i¢i?' where w2 = WiWi.41) becomes V~(¢) ¢iWi(m2 + AW2) + +AWi¢i¢j¢j t [(m2 + AW2)Oij + 2AWiWj] ¢i¢j (3. Note that the masses of the unphysical scalars vanish because we are working in the Landau gauge.
The final result is V V Vv(¢) (3.1.48) 31 .46) £ = . and proceeding as above we find (3.1.j2 3) + 2 3G 2 ( log  log  Wj. since all other Yukawa couplings in the standard model are negligibly small. recalling that a factor gtLv( _gtL + ktLk /k2) = 3 + 2E must now be included because of the form of the vector boson propagators in the Landau gauge. Therefore. (3. we consider only the contribution of the top quark.j2  5) + 6 j.j. and a contribution from the Z boson with mass (g2 + g'2)W2 / 4 and couplings (g2 + g'2)WigtLv/ 4. the oneloop tadpole receives one contribution from a loop of a W vector boson with mass g2w2/4 and couplings g2wig/w/2 to the scalar fields ¢i.j2 G 3)  3Z 2 ( log  j. The result is V(¢) = ~m2¢2 2 + ~A(¢2? 4 ( +2 647r +6W where 1[ 2( H2 H log .36). we must consider the contribution of fermions.4 we find that the relevant term in the shifted lagrangian is (3.1.1.12T 2 ( log  j. The corresponding contributions to the effective potential are easily computed with the help of eq.the vector bosons and the scalarvectorvector vertices needed to compute the oneloop tadpole.44) where again we have chosen Wi = 0 for i #.3)] 2 .3 and W3 = w. the relevant piece of the shifted lagrangian is ht (3. With the help of the results in 4.j2 Z  5) 6 2 . we have computed the oneloop effective potential of the standard model in the MS subtraction scheme. (3. With the choice of w adopted above.47) where we have included a factor of three for the colour quantum number.45) Finally. and a minus sign because of the fermion loop..1. For simplicity.j2 T .1.V2 ( ¢3 + w )tt. To summarize our results.
Symanzik equations In fact. and Ci = 5/6 for vectors.1.54) (3.49) are usually called the field dependent squared masses of the theory.11). (3. On the other hand.48) with respect to log jJ.1. A number of interesting things can be done with the oneloop effective potential (the original work of S. (3.1.52) obey the Callan(3. dV / dt can be computed explicitly by differentiating eq.1.1.57) and therefore (3. and are computable in perturbation theory.1. there is one such function for each particle in the spectrum. the oneparticle irreducible Green's functions (3.53) in eq.1. Weinberg is particularly instructive). We will concentrate on some of them.The quantities defined in eq.2.52) is immediately obtained.2 and neglecting twoloop effects.1. eq. (3.50) with the index i running over all particles in the theory. is the spin of particle i. and I are functions of the coupling constants. From eq. Let us consider for example the dependence on the renormalization scale u. (3. We may denote these functions collectively with the symbol (3.59) 32 .Ci] .55) (3.1. Using eqs.1.1.51) where s.1. We find dV(¢) dt (3.58) (3. "[m. and its value at (P = v2 equals the squared mass of the corresponding particle. Ci = 3/2 for scalars and fermions. (3.. (3.1.1. we have dV(¢) = 0 dt ' where t = log jJ.1. (3. and rewrite the oneloop correction to the scalar potential as V1(¢) = 64 7r2~(_1)28i 1 (2Si + 1)Mt(¢2) [lOg M~~¢2) . Coleman and E.53) where (3.1.11).56) and /3>.
1.64) + . The stability condition can be translated into a lower limit for the Higgs boson mass. but almost so: infact . a necessary condition for the existence of a minimum of V(¢) for finite ¢2.60) 1 +m 2¢2 [ 1 + 2 A2] .j2 or.1.61) is just the expansion of the renormalization group improved effective potential (3.3.2? 8 16 _ 3h4] log A2} t j. ¢2 = ¢2(j.62) (3.63) (3. (3. To see this.1. (3.= 1 dt 167r2 [12A2 + 3 3 l + (l + g.Observe that eqs.56) have the approximate A2 A(A) ~ A + f3).59).65) We see that the stability condition for the potential is simply the positivity of the running coupling constant A(A) at large scales.1.. where A is some energy scale much larger than the electroweak scale.543. (3.3h4 t  3Ag2 __ 3]A(g2 2 + g.m log ~:) ¢2 (A) ~ ¢2 (1 + 2.j).j m (A) ~ m (1 2 2 (3. It is now immediate to show that eq.66) This equation must be solved together with the oneloop renormalization group equations for gauge and Yukawa coupling constants.1.1.We will be interested in discovering under which conditions V (¢) + +00 for large ¢2.j). m2 = m2(j.61) We now observe that the renormalization solutions group equations (3. which in the standard model are given by dg 1 ( 19g dt = 327r2 6 3) 33 (3.2) + 6Ah2 t .j2 (3.1.67) .1.3.1. for example " to obtain the others.1. we need the explicit form of the oneloop renormalization group equation for A(j.59) are not quite enough to compute all the anomalous dimensions of the scalar sector. (3. the effective potential is approximately given by I"V V(¢) ~ ~¢4 {A 4 + _1_ [12A2 167r2 12A log327r2 + ~l + ~(l + g.1.j): dA . using eqs. Under this assumptions. (3. We will now study the behaviour of the effective potential for large values of the classical fields (Pi.1.log2 j.j).log ~:) .1.58. with A = A(j. We therefore assume that ¢2 A2. j. it is sufficient to compute explicitly one of them.2? 8 16 .1.1.58.
where (3. = 175 GeV).69) (3.j = mz). We see that if the initial condition at j.1. We observe that.0 Figure 3.j = v is small. 3. coupling constant ).1 is obtained for m.1.1 in connection 1. and the MS scheme is adopted.1. the coupling constant A grows with increasing u.71) The interpretation of fig. j. The result for A(j.68) (3. translates into a lower limit on A (v)..8 0. 3. This system of coupled firstorder differential equations can be easily solved numerically.100. as ex potential is as follows.j) > 0 in a smaller range of u.1 for different values of the initial condition A(j.130. 3.(mz).4 0.(/l) for different values of ).0 with the problem of the stability of the effective 0. There is another lesson to be learned from fig. or equivalently on mHo This lower bound depends on A..1: The running plained in the text. 3. we see for example that if we ask A(j.1.70) where ss is the strong interaction coupling constant. and eventually leaves the I"V I"V I"V 34 . the Higgs boson mass cannot go below 150 GeV (fig.(3.150.j) is shown in fig. for large values of the Higgs boson mass.j) > 0 up to the grand unification scale.j A. Namely. the requirement that A(j. Conversely.j) becomes negative for some value of the renormalization scale. 1016 GeV.j) stay positive at least up to a given value of u.1. we have chosen A(mz) corresponding to mH = 60.190 and 210 GeV. then A(j.6 3 e 0.2 0. This lower limit becomes less stringent if we require A (j.
for the theory to make sense up to a given scale A. ::. coming from studies and lattice simulations of simplified theories. Both the triviality upper bound and the stability lower bound on the Higgs mass are shown in fig. although with a large uncertainty. we must ask A(jJ. that affects various observables (like the W boson mass.) < 1 (or something like that) for jJ. LEP and SLD precision data allow to estimate. This is because the solution of the renormalization group equation for A has a singularity in jJ" known as the Landau singularity. Also in this case.perturbative domain. we are forced to require the consistency of the theory only up to some finite value of jJ" and to assume that some new phenomena become relevant at higher energy scales. Recent I"V 300 ~ R~ t!J a a 200 t~ 100 Figure 3. the allowed range for mH becomes narrower. there are only some indications of this. = 175 GeV. It is interesting to notice that a value of ttiu in this range is compatible with A close to the unification scale. As A increases. A. The central values of these fits are between 100 and 200 GeV. as functions of A. Therefore. 1016 GeV. This in turns implies an upper bound on the Higgs boson mass. The reason for this is that the existence of a Landau singularity in the running coupling constant A would imply A (v) = 0 if we require that the theory be valid for all values of the scale jJ" that is. Notice however that no rigorous proof of the triviality of the standard model has been given so far. or forwardbackward asymmetries) through radiative corrections. which is approximately 180 GeV for A 1016 GeV and m. 3.2: Theoretical upper and lower bounds on the Higgs mass. A < 1. the value of the standard model Higgs mass. in the scalar sector. the theory would be noninteracting. The upper limit on the standard model Higgs boson mass is often referred to as the triviality limit.2. or trivial. I"V 35 .
1) could in principle be modified at higher orders in perturbation the measured value of p is very close to 1: Pexp = 1.5) that is.2. Therefore.2. tan()w = Jm~.4) This in turn implies that (3.2.2) Equation (3..9) 36 . Next we notice that the scalar potential (3.2. thus suggesting that radiative corrections. p = 1 only if M2 = m~..2.8) can be interpreted as the squared length of a real four vector. We define a 2 x 2 matrix (3. the scalar potential has an 0(4) SU(2) x SU(2) invariance.2 ( W3tt ) B· tt (3. In fact. Actually. Therefore. M. even after the inclusion of radiative corrections.2. say M2.3. we general vector boson Lmass  some symmetry property prevents the quantity P from receiving large We will now show that this is indeed the case.2.2. Preliminarly.1) theory. the most mass term is given by _1 "2mw W 2( 1tt 1 2tt 2) 1( tt tt) [M2 Wtt + W Wtt +"2 W3 B M'2 MI2] M.M2 M (3. defining q) = ~ we see that which is larger than the standard model SU(2h x ( :: : ~:: ) (3.6) is invariant under a group of transformations U (1 )y.2 The SU(2) custodial symmetry model at tree level the weak vector boson masses We have seen in section 3 that in the standard mw and mz are related by P m2 m2z cos? ()w w = 1.2.3) the condition that the photon stays massless gives us M'2 = M M".7) 1 q) 12= I"V t(q)i + q)~ + q)~ + q)~) (3. and M2 + the mass matrix in the neutral sector is completely fixed by the value of one parameter. This symmetry property can be implemented in the following way. and it is diagonalized by a rotation of an angle ()w given by Furthermore.0022.2 = m.0048 ± 0. (3.. observe that.2. (3.
2. however.¢f transforms as an SU(2) the action of a generic SU(2)L transformation U. where V is a second SU(2) constant matrix. Let us therefore neglect for the moment the hypercharge factor of the gauge group. in order to work with an 0(4)invariant theory.2.12) H + UHVt.12) is a representation of the 0(4) symmetry we mentioned above.2. since radiative corrections to p due to the hypercharge coupling are very small.2. More precisely.3). We are almost at the end of the road: in fact. we have H + UH.1) of the order of GttmJ. (3. Of course.13) is not equal to (Dtt¢)tDtt¢ (prove this statement as an exercise). However. We conclude that the 0 (4) symmetry is violated by the hypercharge interaction term contained in the covariant derivative. Due to spontaneous breaking of SU(2)L. (3.9) is preserved also by right multiplication with an SU(2) matrix.2. one finds p~1+ where we have included only the contribution 3G m2 87r 2tt t ' v'2 (3. it follows that.2. it is easy to check that the only mass term for the W~ fields allowed by the symmetry in eq. the ground state is not invariant under 0(4). We have proven that p = 1 is a consequence of the socalled custodial SU(2) symmetry defined in eq.2. doublet.15) from the top quark. Is it possible to write also the kinetic term for the field ¢ in an 0(4)invariant way? The natural candidate is of course (3. (3. (3.14) is of the form w~wt. for obvious reasons.2.14) that leave the vacuum expectation value < H >= v'2VTl unchanged (U is now xindependent).2.13) which is invariant under the transformations (3. Equation (3. it is easy to check that the scalar potential can be written in terms of H as (3. which amounts to setting g' = 0.2.Recalling that the field ¢c = (¢o*. M2 = m~ in the notation of eq. independent of U.2. this is because ¢ and ¢c have opposite values of the hypercharge quantum number. fermion mass terms do not preserve the custodial symmetry. under (3.2. one readily realizes that (3.2. we expect corrections to eq.that is.12) since D" + iitro'. This is possible because the structure of H in eq.11) which is invariant under the SU(2)L x SU(2) transformation (3. (3. and therefore it is not spoiled by radiative corrections.2. In other words. The inclusion of the hypercharge interaction.14). there is a residual 0(3) SU(2) symmetry under transformations of the kind I"V (3. a scalar product in 0(3).10) On the other hand. that breaks 0(4) explicitly. 37 . does not change this conclusion.
Exact axialcurrent conservation is obviously recovered in the m + 0 limit. and that must hold at all perturbative orders for the theory to be renormalizable.5) is well known. (3. It is easy to show. that are consequences of current conservation. using the equations of motion.3. gauge invariance manifests itself in the form of identities between Green functions.tt'ljJ (3.6) The lagrangian of massive QED is not invariant under axial transformations because of the presence of the mass term.3 Axial anomaly cancellation We have seen in the previous sections that the renormalizability of the standard model is strictly connected with gauge invariance.4) and (3. 'ljJ with electric charge e and mass m. Equation (3.3) J~ = 'ljJ. and we will then state under which conditions the standard model is anomalyfree and renormalizable. we have seen that the massive vector boson propagators show unphysical singularities. The current J~.3.3.2) (3.4) (3. that are cancelled by the presence of wouldbe Goldstone bosons.8) 38 . as the standard model itself. In turn.Taylor identities are called anomalies. and as a consequence the associated current J~ is not conserved. called Slavnov.3. Equation (3.4) is simply the conservation of the electromagnetic current. It might happen that current conservation is spoiled at the quantum level.3. We will illustrate the problem of anomalies in the context of a simple example. In the language of quantum field theory. (3.3.3. on the other hand.3. unless the spectrum of the theory satisfies particular conditions.7) which can be easily shown to be related to the matrix element of the axial current between the vacuum state and a twophoton state by the relation (3. In particular.5) precisely states this fact.tt'5'ljJ Jp = 'ljJ'5'ljJ. we will show that this might not be the case for theories with axial currents. Now consider the Green function (3. We consider the operators Jt = 'ljJ.5) The interpretation of eqs. is associated with axial transformations. We consider quantum electrodynamics with one massive fermion.Taylor identities. (3.3. terms that spoil the validity of Slavnov. which reflects the gaugeinvariance of the theory. In this section.1) (3. that 8ttJt = 0 8ttJ~ = 2imJp .3.3.3.3.
3.3.5). T/LV(kl' where k2) = TiV(kl' k2) k2) + TfVP(k1. 3. (3. k2) = TiVP(k1. and the canonical commutation relations.17) Tfv (kl' k2) = T~/L(k2' kl). 39 .3. We have T/LVP(k1.3.3. k2).3.14) (3. where q = kl (3. the diagrams to be computed are those of fig.3. (3. it obeys the Slavnov.9.3. k2) (3.3.9) (3.3.3.9.3.3.3.12) (3.10) can be worked out by exploiting eqs.11) The identities in eqs.3.10) + k2 and T/LV (kl' k2) = i f d4xld4x2eihxl +ik2X2 (OIT[ Jt(Xl)JV(X2)Jp(0)lI0) .4) and (3. (3.3.3. We will now check explicitly whether eqs. At the oneloop order.16) (3.Formally. k2) = T~/LP(k2' k1) (3. (3.3.15) and TfVP(k1. (3.10) are satisfied in perturbation theory or not.13) + TfV(k1.Taylor identities kiT/LVp = k~T/Lvp = 0 qPT/Lvp = 2mT/Lv.
they cancel against the fourth and second terms. (3. for finite M.18) in T'wP .3. that it is possible.21 ) and also (3. The regularization scheme we choose is the following. We subtract from each integrand in eqs.3. (3.12) and (3.14) and (3.20) Now.19) in Tlj'vp (and similarly in the regularizing part of the integrands). respectively.9). (3. although quite complicated. we find (3. thus obtaining announced results.3. We have therefore (3.3.3. (3. the integrals are now convergent.15) are superficially divergent.14) and (3.22) by an analogous argument. In the limit M + 00 the original expression is recovered.3. but with m replaced by an arbitrary regularization parameter M.3. because of the presence of /5. shifting k + k + k2 in the first term and shifting k + k . that state the conservation of the vector current. writing (3. We must therefore choose a regularization scheme before proceeding. Dimensional regularization is not suited here.3.3. The limit M + 00 can then be taken safely. while. Using the (3. are satisfied by T'WP as given in eqs. We will indicate with a subscript M the regularized quantities. which has an intrinsically fourdimensional meaning and cannot be generalized to other spacetime dimensions in a simple way.3. Equations (3.kl + k2 in the second one.3. keeping in mind.3. The loop integrals in eqs. however.15) the same expression.3.The overall minus sign is due to the presence of a fermion loop. We will make a different choice.3.14). In fact. We may use a similar procedure to check the identity in eq.10).23) 40 . to treat this problem within dimensional regularization. and 1 (3.
dan n (3. three and five I matrices.m i +~+~2m'51 i v i ~ .24) in qpTiVP we get where RI"V = and qpT. The important point here is that these shifts in the integration variable can be performed only after regularizing the integrals..27) Let us now compute [2mTI"V]M 1 explicitly.3.and (3. Therefore.(m + ).30) with the result (3. they cancel against the third and the first respectively. The simple expression in the numerator is obtained by dropping all products of 15with two.28) we find [2m TI"v 1 ]M =2 0 10 1 d 101X x 0 d Y f dk [ (27r)4 [k2 4 CT kPk ." The integration over the loop momentum k can be easily performed by shifting the integration variable (3. by shifting the loop momentum k to k + k2 in the second term.3. and to k + k1 in the fourth. and exploiting the antisymmetry of EpVCT!. (3.mi· v] (3.:vp respectively (and making similar replacements in the terms with m + M).3.31 ) 41 ..26) It is now easy to see that [RI"V]M vanishes..3.m I . Using the Feynman parametrization d1 a1 .~ ~1   I" i m 151~.k Tr [i (27r)4 ~+ 4 ~1  m v 151 I" i I I" ~.3.29) where we have set ki = ki = o.m ~ i ~2  m v 151 i I I" ~. (3.8· 2 EI"VPCT1 2 im M] + 2k(k1x _ k2Y) _ m2]3 .3. In fact. (3.3.3.25) f d .
which are divergent. fermions are either in the doublet or in the singlet representation of SU(2). even ifm = O. The anomalous term can be taken into account by modifying eq.Taylor identity in eq.3. n with masses ttu. 1 EI"Ypu kPku 2· (3. vector charges Qi and axial charges Qf: aI" JI" A where now  ~ ~ Q5Q2 i i [2· 'lmi Jip + (47r )2 Etu/pa 1 r I"Y pa D F 1 .34) J~ = L Qf'l/J(Yl"i5'l/Ji. (3.. i = 1.3.. (3.3. Notice in fact that the anomaly is independent of the fermion mass.3.3. The first quantity is obviously zero: (3. Tb}TC) Tr ( {r".3. (3. (3.39) (3.35) The above considerations can be extended to the case of a theory with nonabelian gauge invariance.3.36) where T" are the gauge group generators. a regularization procedure is needed in order to prove the cancellation of integrals with two propagators. i=l n J~ = 'l/J(Y5'l/Ji .40) must all vanish.3. or the AdlerBardeenJackiw anomaly.33) can be immediately generalized to a theory with n fermion fields 'l/Ji.3. for the axial anomaly to be cancelled. the axial current is not conserved. this means that the four quantities Tr ( {r".3. (3.3. The limit can now be taken safely. This term arises because of the impossibility of regularizing the theory in a way that preserves both the vector and axial vector classical current divergence relations. In this case. which is usually called the axial anomaly.32) 1 27r The effect of the regularization is that the Slavnov.33) where FI"Y is the fieldstrength tensor of QED.Notice that the RHS of eq. also fermion loops with four and five internal lines contribute to the anomaly.37) (3. however. In the standard model.41 ) 42 . at the quantum level.3. In other words. . it can be proved that higherorder corrections do not modify the oneloop expression of the anomaly. Tb} Y) Tr (y2TC) Tr (y3) (3. . Furthermore. The anomalous term is finite.3.10) is spoiled by an anomalous term.5) at the oneloop level in the following way: 2m T I"Y  2 I" aI" JA  2· im p J + (47r)2EI"YPU 1 FI"Y FPU .38) (3. one of the two must be given up.3. (3. It can be shown that the anomalous term of the axial vector current in a nonabelian theory is proportional to (3. (3.31) is finite when M + giving q PT /wP  00. The result in eq.
and Y = 1 for the doublets of lefthanded leptons. To show this. the currents associated with the leptonic and baryonic numbers.46) (3. we have simply TC = 0).48) It is easy to prove that.nl).43) where nq (nl) is the number of quark (lepton) families. and the overall factor of 2 is due to the fact that lefthanded fermions are SU(2) doublets.3. which has been recently confirmed by the discovery of the top quark.50) 43 .42) where YL is the hypercharge matrix restricted to lefthanded fermions. Tr (y2TC) = 0. The factor of 3 in front of the quark term is due to the colour degree of freedom. it is convenient to write the axial current as (3.47) Tr (Y~) = 3nq [ (~) 3 + and therefore (_~) 3] + nl( _2)3. Since t" = 0 for righthanded fermions. Tr (y3) = 0.49) (3. We see that the cancellation of the axial anomaly requires that the numbers of quark and lepton families are equal! This is an important prediction of the standard model. since Y has the same value for both components of each doublet. it is clear that lefthanded fermions and righthanded axial anomaly with opposite signs. The last condition. we find Tr (YL) = nq x 3 x 2 x "3 + nl x 2 x 1 (1) = 2(nq . (3. Since Y = 1/3 for the doublets of lefthanded quarks. is also satisfied provided nq = tu.3.44) In this way.3. The third condition. We have therefore fermions contribute to the (3. we have (3. is again trivially satisfied.T3) we find Tr (Yt) = 6nq (~) 3 + 2nl( _1)3 (3.3. because of the axial anomaly.3. (3. and Tr (TC) = 0 (for singlets.The second quantity requires more care.3.3. V" = L [ei/I"ei + i=l nl Vi/I"Vi] (3.3.45) Using Y = 2(Q .3.
(3.53) We now consider triangle diagrams with L~ or L~ on one vertex. and therefore aI" LI"  1 (47r)2E I"VPeT[. (3.3. The difference B . we get a LI"  I" R (47r)2E 1 I"vpeT.2 9 B I"V B pa T {YR r 2' 2 . we may write for the baryonic current (3.59) where (3.3. due to instanton effects. eLhl" ( ~~ ) L~ = eR/l"eR. Indeed.54) Thus.3.2B 9 I"V B pa  9 2Wi I"V Wi pa ] .58) This results in a (numerically negligible) nonconservation of leptonic and baryonic numbers L and B.57) since YR = 2.are anomalous.61 ) 44 . and weak vector bosons on the the two remaining vertices. (3.3.51 ) L~ = (ih.55) The minus sign arises because /5 appears in L~ with a minus sign.3. Clearly.(47r)2E YR} _ 2 I"vpeT.L is however conserved.56) By a similar argument. and let us rewrite the leptonic current as V" = V" + V" L R' where (3. Using YL = 1 and the anticommutation relations among the Pauli matrices.3.2B 9 I"V B pa + 9 2Wi I"V Wi pa ] .3.3.3. let us consider the case of only one generation (the extension to more than one generation is trivial). (3. (3.3. In order to prove this statement.3. only lefthanded (righthanded) fermions circulate in the loop with L~ (L~).2 9 B I"V B PeT.52) (3. This is easily seen by working out the Dirac structure of the loop integrand: (3.60) (3. we find aI" LI"L(47r)2E1 I"VPeT [.
4.2) Here.63) we get 1 /LVPeT[ 12 B B (47r)2E 9 /LV peTg 2Wi Wi ] /LV pa (3. j +_1_E/LVPeTg.64) which is exactly equal to a/LL/L.and compute a/LB/L as in the case of the leptonic current. the lagrangian variant under a large class of global transformations: namely. 2'2 The global factor of 1/3 is cancelled by a factor of 3 from color. Using the known values of quark hypercharges (3.2 (47r)2 pa [Tr {YuR YUR} 2'2 + Tr {YdR YdR}]. 1) (1.3.4 Accidental symmetries The need for a Yukawa interaction term of fermion fields with scalar fields can be motivated in a different way.1) is manifestly inthe fermion fields within each (3. Consider the standard model with only one generation of quarks and leptons. the symbol+.. 3. in eq.3. In addition to the assumed gauge symmetry. and the two numbers representation (2 for the doublet.4. and no scalar fields.1/3). 2/3) (2.4. (3. 2) =L = dR =Q = UR (2.4/3) (1. representation can be multiplied by an arbitrary constant phase 45 . Mass terms are forbidden by the gauge symmetry. in brackets stand for the SU(2) hypercharge quantum number. 1 for the scalar) and for the respectively.means "transforms as". aHB/L r We find =  1 E/LVPeT[gl2 B B (47r)2 /LV B B /LV pa Tr {Y 2 2' Y Q Q} + g2Wi /LV wj pa Tr {Ti 2' 2 T }] (3362) . The lagrangian for fermion fields can be written in the following compact form: (3. This shows that the current B/L .1) k=l where the sum runs over the five different irreducible representations fermions in a generation: of SU(2h ® U(1)y of the 'l/Jl 'l/J2 'l/J3 'l/J4 'l/J5 = eR rv rv rv rv rv (1.L/L is conserved.
they are incompatible with mass terms.6 .. The Yukawa interaction terms of the previous subsection break this larger accidental symmetry too.iLu + d.3) .4. Let us take a closer look to the accidental symmetry.iL'5e + Jf + Jt) Jt 3 (u.iLeR Jf = ih. On the other hand. rather than [U(1)]5. This [U(1)]5 global symmetry was not imposed at the beginning: it is just a consequence of the assumed gauge symmetry and of the renormalizability condition.iLUR Jf = dR.2) are Ji = eR..without affecting L. Consider for Jr. which corresponds to a local invariance of the theory. experiments show no sign of the conservation of J£5 and Jb5. 1025 Y . let us briefly review the most important experimental evidences of baryon and lepton number conservation. respectively.iLv + e. The most obvious test of baryon number conservation is proton stability.iL'5U + d.iLdR Jt = UL.iLdL Equivalently. because of flavour mixing.. Individualleptonic numbers are separately conserved. leaving however baryonic and leptonic numbers conserved. while only the total baryonic number is conserved. the accidental symmetry gets much larger. To conclude this subsection.iL'5V + e. The five conserved currents corresponding to the global transformations (3.iL'5d. The experimental lower bound on the proton lifetime is at present Tp > 1.iLd) U. When the theory is extended to include more fermion generations. In fact.iLVL + eL. 46 (3.iLe V. The current Jv is the hypercharge current. since also mixing among different generation is allowed. in a realistic theory.iLUL + dL.iLeL Jf = UR. in such a way that the Jt. = k=l i 11 Yk Jt: 2 J% = Ji + J%5= Ji Jt = 3(Jf Jt:5 = Jf + Jf Jf Jf  v. since baryonic and leptonic number are known to be conserved to an extremely high accuracy. The invariance of the lagrangian under the corresponding global symmetries is certainly good news.4. the corresponding symmetries should be broken. It is therefore called an accidental symmetry. one could define the accidental symmetry transformations corresponding currents are five independent linear combinations of example the choice »: . and they are broken by the Yukawa interaction terms that generate fermion masses via the Higgs mechanism. The currents J£ and Jb are immediately recognized to be the leptonic and baryonic number currents. The true accidental symmetry is therefore [U(1)]4.
4. + 3e) ::. 47 . 1.6) (3.The most accurate tests of lepton number conservation are provided by the following observables: + e.) ::.) ::.4.Ti r(j).1012 B(j).1011 B(j).7.4.Ti B(T (3.106. 1.5) (3.2.7) + e Ti) < 4.. r(j). 2. 1012 + all)  + j).4.4) (3.
AddisonWesley from isospin to unified theories. (1995).B. Golowich and B.V.F.R. Holstein. Peskin. Li. Lie algebras in particle physics: Cummings (1982).F Donoghue. Clarendon Press to gauge field theory. Adam Hilger (1986). [4] L. The quantum theory of fields. [5] T. Okun. [7] J. Leptons and quarks. to quantum field theory. [8] S. Weinberg. An introduction (1981). Field theory: a modern primer. BenjaminCummings [2] M.Bibliography [1] P. Cheng and L. Love. North Holland (1982).E. Schroder. Ramond. Gauge theory of elementary (1984). 48 . Georgi. Benjamin [3] H. Bailin and A. Introduction particle physics. D. Cambridge University Press (1992).P. E. [6] D. Cambridge University Press (1995). Dynamics of the Standard Model.
1) where Vi is the number of vertices of type i. Therefore. we have (4. each type being labelled by the index i.1 Renormalizability and power counting In this appendix. nL di be the number of fermionic lines.1. which in turn is equal to the total number of internal lines I = If + Ib minus the number of independent momentum conservation equations.h internal bosonic lines.V vertices. . let n}.E f external fermionic lines.Eb external bosonic lines. The following relations hold: (4. we describe the powercounting criterion for renormalizability of local field theories. .1.1.2) (4. 49 .4) We now define the degree of superficial divergence D of the diagram as the power of momenta in the numerator minus the power of momenta in the denominator of the Feynman diagram. .1.Chapter 4 Appendices 4. . and that (4. Consider a Feynman diagram containing .If internal fermionic lines. respectively. Finally.L loops.3) The number L of loops is equal to the number of independent internal momenta. Let us assume that there are different types of vertices. bosonic lines and field derivatives in typei vertices. .
the condition in eq.7) ° ° 2 f 2 b and it must hold for each i separately (a diagram can contain only vertices of one type). and the whole theory can be made finite by renormalizing only these primitively divergent amplitudes.1.1.s. Notice that the 1. of eq.1. corresponds to one power of momentum. each field derivative and d powers of momentum are carried by each loop Now.2Ib + L diVi. (4. at any order in perturbation theory. 50 .5) and (4. The condition for renormalizability then becomes d1o d2 dZ + __ nz + __ nz < d (4. D = dL .h. then only a finite number of diagrams have D ~ 0.1. the Feynman amplitude will be ultraviolet divergent. if the last term in the r. we find d1o + "'vz 0(0 d + __ c: 2 Z Z d2 Eb 2 nz f d2 + __ 2 nz ° b  d ) . of eq. Therefore. (4.5) since fermion propagators behave as k:'. However.6) is zero or negative.1.1. (4.1.Clearly. On the other hand. (4.7) is just the mass dimension of the operator that corresponds to type i vertices: in fact.1.h. and eliminating If and h via eqs.1.1.If .2) d1 D = d .4) in eq.s.1. For this reason.3). fermion fields have dimension 3/2. D < 0 is not a sufficient condition for convergence. integration in ddimensional spacetime.Ef 2  boson propagator behave as k:".1) and (4. replacing eqs. boson fields have dimension 1 and derivatives have dimension 1.7) can be rephrased in terms of coupling constant dimensionality: a renormalizable theory can contain only constants with mass dimension ~ o. we notice that D decreases with increasing number of external lines. (4. i (4. (4. since there can still be subdiagrams with D ~ o.6) If D ~ 0. (4.
(4.2 Nonunitarity of the Fermi theory In this Appendix we will work out the restrictions imposed on scattering amplitudes by the unitarity condition of the scattering matrix. We get or f i L II f (27r) d 3 ~/ 2Ei f (27r)4 0(4) (Pa  L p/) i MbfM:f = i (Mba . and the amplitudes 51 .2. with momenta PI.5) where p/ is the momentum of particle i in the state f. Writing the scattering matrix as S = I + iT.2.2. the states If) are also twoparticle states of the same species as those in la). by (4. b we have (aITtTlb) = i ((aITlb) .4) and insert the identity operator between Tt and T in the 1. the unitarity condition (4.2) For generic states a.h.2.2. Under these conditions. eq. furthermore.7) For a = b. (4. (4.4.s.M:b) .8) which is the socalled optical theorem: the total cross section for the process a + f is proportional to the imaginary part of the forward invariant amplitude Maa.7) gives (4.(aITtlb)) .2. P2.2.2. and we will show that the Fermi theory violates this unitarity bound at sufficiently high energy.3): (4.2. let us assume that only elastic scattering is allowed.1) st S = I gives (4.3) Now define the invariant amplitude Maf for the process a + f (4. Let us now assume that la) is a state of two particles of the same species. of eq.
9) (4.11) where is the scattering angle. is given by 2ImM(s. (4. for a given value of the centerof mass squared energy s.2.2. (4. t) is a function of cos e only. )J (4.MaJ depend on the initial and final states through the two independent Mandelstam variables s.s.2. t) where the partialwave amplitudes = 167rL(2J J + 1) aJ(s) PJ(cos e).8) we get f d3Pl d3P2 4 (4) 2 (27r)3 2El (27r)3 2E2 (27r) r5 (PI + P2 . of eq.2.16) Replacing eq.2. (4.2. Thus.s.10) In the centerofmass frame.15) in the 1.18) 52 .2. t: MaJ where M(s.cos 2 s e) + cose =1 +. (4. the amplitude M(s.h.PI .2. (4.13) and the normalization We find conditions (4. and can be expanded on the basis of the Legendre polynomials e PJ (Z ) = J' . t = (1.2.14) M(s. t).h.2. t)1 _1_ 167r /1 dcos e 1 J [167r L(2J J + 1) aJ(s) PJ(cos e)] [167r L(2K K + 1) a'k(s) PK(cos e)] 327rL(2J+1) laJ(s)12.2. s 2t (4.0) = 327rL(2J J + 1)ImaJ(s).17) while the r.P2) IM(s.2.2 J d ZJ 1 d J (2 Z  1. (4.12) The Legendre polynomials obey the orthogonality conditions (4. t).15) aJ are given by = 1 aJ(s) 327r /1 d 1 cos e PJ(cos e) M(s. (4.
(4. (4. as appropriate for a forward amplitude. (4.~ 875 GeV.14).27) 53 .2.a(1'5)P2'i3(1 '5) ~d (4.21 ) within the Fermi theory.23) by performing the replacement  GF V2 + g2 c: 1 8t m?v· (4.23) 32G1s2.a(1'5)ih .20) Let us now consider a specific process.: (4. the numerical factor is conventional).2.2.19) laJ(s)12 = ImaJ(s) for all partial amplitudes. t)12 G.2. The relevant amplitude is (4. or equivalently cos () = 1.. (4.2. We see that only the partial amplitude ao (s) is nonzero.25) = . unitarity of the scattering matrix requires (4.2. Tr [.16) we obtain lao (s ) I = The unitarity GFs 2v27r rs.2.19) provides the unitarity bound (4. where a sum over polarizations is understood.22) which gives IM(s.2. Using the definition eq.2.2. Therefore. namely the scattering (4.26) Let us now repeat the same calculation in the context of a theory with an interacting vector boson W with mass mw and coupling g/(2V2) to lefthanded fermions (the coupling 9 is dimensionless. since there is no t dependence at all. and we have used the normalization condition (4. Equation (4.20) is therefore violated at I2A VB = ~ G.24) bound eq. G~s 27r (4.2.2.2. The squared amplitude in this theory is obtained from the result in eq.2.13(1 '5) ~2]Tr [.2.23) we obtain the total cross section (J (4. From eq.where we have set t = 0.
mw (4..2.31 ) The value ofmw is related to the size of the coupling 9 through eq. this expression reduces to the result obtained in the Fermi theory.2. 8 m'iv .2. At very (4.28) The total cross section is now given by (J =  g4 647rm?v 8 + . about 900 GeV.j2 1 )2 2 g48 2 8 =( 2)2· 8 t . the linear growth of the cross section with large energy we have is cut off at 8 I"V m~. (4.7. (4. in this case we get 9 0..2.mw t .30). Ifmw were close to the energy at which the Fermi theory breaks down.29) For 8 « m~. I"V 54 .30) 8 w In this case. (4. The fact that the measured value mw is instead much smaller.2. (4.We get IM(8.26). mw ~ 80 GeV. however. eq. then 9 would take a value close to 10. with the identification 8m2.2. is a signal of the fact that a theory of weak interactions with an intermediate vector boson can be treated perturbatively: indeed.t)1 2 =32 2 (g2. far from the perturbative domain.
3.2) where a is a real constant. (4. The transformation that This gives (8IL .3.3) The ordinary derivative must be replaced by a covariant derivative.ieA'IL)eiea'ljJ = eiea(8IL .2) with a = a(x). that is. in order to make it invariant under the field transformation (4. The constant e plays the role of the conserved charge associated with this invariance property.3.3.3. (4.ieAIL)'ljJ (8IL . we want to modify £.ieA'IL'ljJ = 8IL'ljJ .ieA'IL)'ljJ' = eiea(8IL .3. .3 Gauge theories The abelian case The Dirac free lagrangian for a massive fermion E = 'ljJ(i8 . AIL + A'IL = AIL 55 (4.4) where AIL is a real vector field.6) property of AIL must be fixed in such a way (4. .m)'ljJ is invariant under the local (or second kind) gauge transformation 'ljJ + 'ljJ' = eiea(x)'ljJ 'ljJ + 'ljJ' = eiea(x)'ljJ. The derivative term is not invariant: (4.AIL'ljJ which implies (4.ieAIL)'ljJ 8IL'ljJ + ie(8ILa)'ljJ .9) . (4.ieAIL'ljJ (8ILa)'ljJ .3.m)'ljJ is invariant under the global (or first kind) U(1) gauge transformation 'ljJ + 'ljJ' = eiea'ljJ  (4.7) The lagrangian E = 'ljJ(iD .1) 'ljJ + 'ljJ = ezea'ljJ.3.8) (4.3.4.3.5) + 8ILa(x).A'IL'ljJ = . We want to promote this global symmetry to a local one.
A kinetic term. and the vector field AM is now a hermitian matrix AM = A~tA.3.3. must now be introduced.3.3. The covariant (4. For definiteness.3.3. The nonabelian case Let us consider now the case when the invariance group of the theory is nonabelian.18) law (4..13) can be expressed in where jABC is completely antisymmetric.B. . It is given by (4.. we observe that no selfinteraction vector field AM is present in the lagrangian. (4. This is connected with the abelian invariance group. (4.10) where (4. It is uniquely fixed by the requirements of Lorentz and gauge invariance..11) Notice that (4. Finally.14) 9 in analogy with the abelian case. This group has N2 .1 hermitian traceless generators t/'.3. where we have inserted a coupling constant derivative is now given by (4.3.Notice that the requirement of local gauge invariance generates the interaction term e'ljJ'M'ljJAM. A generic element U of SU(N) A terms of the generators t and of a set of real functions oA (x) by U1 = ut .C=1. involving derivatives of the vector field AM.16) It is easy to show. and by assuming the standard normalization of the propagator for AM.12) and that FMV the presence term for the nature of the is invariant under a gauge transformation.17) + !_U(8MU1) 9 56 .3.N21. Notice also that gauge invariance forbids of a mass term for the gauge field AM. in analogy with the abelian case. that the transformation AM + A'M = U AMU1 ensures that (4.15) where I is the unity matrix in the representation space. that obey the commutation relations A. we consider the group SU(N) of N x N unitary matrices with unit determinant.
Consider now an infinitesimal gauge transformation (4.3.19) To first order in a, eq. (4.3.17) becomes
AIL AIL]  '!_igaILaAtA 9 A~tC  gaA A~fABCtC + aILaCtC,
+ ig[aAtA,
(4.3.20)
or (4.3.21 ) A kinetic term for the gauge fields can be built in analogy with the abelian case. Recalling eq. (4.3.12), we define a field tensor FILV through We have
(4.3.22) where 'ljJ is a multiplet of some SU(N) representation, and FILV = F!:ttA. We find
(4.3.23) The kinetic term is then given by (4.3.24) In the nonabelian case, selfinteraction terms among the gauge fields are present. This is related to the fact that, contrary to the abelian case, the field strength FILV transforms nontrivially under a gauge transformation: (4.3.25) For an infinitesimal gauge transformation, we find (4.3.26) which means that the components gauge group.
F!:t
form a multiplet
in the adjoint representation
of the
57
4.4
The standard model lagrangian in renormalizable gauges
model lagrangian: (4.4.1)
Let us consider the following part of the standard
where (4.4.2) (4.4.3) (4.4.4) For the moment, we do not specify the value of the hypercharge quantum number Y of the Higgs doublet ¢. We define (4.4.5) where (4.4.6) and
VI, V2
are arbitrary
complex numbers, only restricted 1VI 1 2
by the minimization
2
condition (4.4.7)
+
1V2 12 = V 2
m = T·
We have
LD
[att¢t + ~¢t (gWiTi + g'YBtt)]
Lq,q,
[att¢  ~ (gW~Tj + g'YBtt) ¢]
(4.4.8)
+ Lq,q,vV + Lq,q,v.
The first term is simply the kinetic term for ¢, (4.4.9) Next, we consider the ¢¢VV term:
(4.4.10) Equation (4.4.10) contains a mass term for the vector fields, that can be isolated by replacing ¢ with ¢l: (4.4.11) 58
where M2  ~ [g2¢1¢10ij gg'Y¢1T ¢1  4 gg'Y ¢1 Tj ¢1 g,2y2¢1 ¢1 Observe that the matrix M2 has zero determinant:
i
1
.
(4.4.12)
(4.4.13) which vanishes because of the identity (4.4.14) In other words, with only one scalar doublet of any hypercharge, one of the four physical vector boson has always zero mass. This is because it is always possible to find a U(1) subgroup of the gauge group which leaves the vacuum expectation value ¢1 invariant. Let us now diagonalize M2. This is easily done by choosing VI = 0, V2 = V, which is allowed because all the degenerate vacuum configurations are connected by gauge transformations. We find
)..,mass
r
_1

4; 9 V
22
W+ttw
tt
1 + SV 2(Wtt 3
Btt) [
g2 _gg'Y
gg'Y g,2y2
1 (W3tt · B
tt
)
(4.4.15)
The first term is already in diagonal form, and tells us that the charged vector bosons (4.4.16) are mass eigenstates, with masses m~ =
4;lv2.
1
(4.4.17)
The second term in eq. (4.4.15) is diagonalized by the rotation
( Btt
Wf ) _ [ cos ()

sin ()  sin () cos ()
1(
Att'
z» ) .
g'Y tan f =,
9
(4.4.18)
where the combination Att corresponds to the zeromass vector boson. We see immediately that, for Y = 1, Att is precisely equal to the photon field coupled to the electromagnetic current, and () ()w. The eigenvalue corresponding to Z" is (4.4.19) In terms of W:' Att and Ztt eq. (4.4.10) becomes (mw
L<jJ<jJvV = W+ttw;
+
~9H)2 2
+
~zttZtt (mz 2 59
+~
g() H)2 2 cos w
22) and we remain with + LGF = ~gW: +~gW.t~j(¢)j(¢).g'sinOw)Zttj . (4.21~(8ttBtt)2 .cm2 G+G.Z" tan Ow) [GW:(H The third term in LD must be considered in conjunction have .iG)] .~g'Btt [(8tt¢2)t¢2 ..24) the mixing between vector bosons and scalars disappears.4.q. (H . Lq.v + LGF = ~gW~ [(8tt¢2)tTi¢2 ._ ~Cm2G2.4.v = ~gW~ ~gW~ [(8tt¢2)tTi¢1 .H8ttG) (G+8ttG. Adding LGF. Lq. .¢t¢2) + j(¢)] 21~(8ttW~)2 . (4.q. +8ttW~ [~9(¢~Ti¢1 . .~g'Btt [(8tt¢2)t¢1 .23) (4.¢t8tt¢2] [(8tt¢2)tTi¢2 .4.1 1 1 2 +_g2W+ttW(G+G+ _G2) + . we can integrate by parts the first row.V .20 (Att sin 20w + Z" cos 20w?G+G4 cos w +gsinOw(mwAtt .¢tTi¢2) + ji(¢)] +8ttBtt [~g'(¢~¢1 .¢~Ti8tt¢2] .¢tTi8tt¢2] .4.) + iG) + G+W.4.¢~8tt¢2] . Lq.~g'Btt [(8tt¢2)t¢2 . 2~ tt <" w 2 <" z (4.9 zttz G2 2 tt 2 8 cos? Ow tt 1 g2 +.21 ) Exploiting the fact that 8tt¢1 = 0. we find . (4.t~ji(¢)ji(¢) With the choices (4.iG)] .¢~Ti8tt¢2] .G8tt(H + iG)] cvrc:  G+8tt(H .¢~8tt¢2] .25) 60 .G8ttG+) 'l "2 [2gsinOwAtt + (gcosOw t(9COSOW _~(8ttWi)2 2~ + g'sinOw)Ztt(G8ttH tt _ ~(8tt B )2 . . [(H + iG)8ttG[(H  .q.20) with the gaugefixing term.mzZtt sin Ow) (GW: + G+W. We +tg2 sin Ow(Att .4.
.j2 (and £~Pt (4.28) = .2. we find (4..4. These masses vanish in the Landau gauge.j2(v + H)ehLe .4. The last term to be considered is the scalar potential V (¢).G.. (2. as is necessary in order to cancel the unphysical singularities in the vector boson propagators.27) We consider now the interaction between fermions and scalars. we get £~adT = G+ (7fhVhDdR . .26) where (4.j2(v + H) (dhDd + uhuu ) .2. 1 (4..uhU/5u ) .dLVthuUR) 1 iG dhD/5d .. From eqs.G+7JhLeR .(dRhDVtUL .4. After some algebra.432.uRhuVdL) . respectively.29) where sums over generation indices are understood.. (2.46) and the definition in eq.52). ~ = o.4.GeRhLv. 61 .2.We see that the wouldbe Goldstone bosons G± and G have acquired squared masses equal to ~m~ and ~m.
Divergences will therefore appear as poles in d . (4.5.m2)2 is logarithmically divergent at large momenta in fourdimensional spacetime.5. for example.5.5. like those appearing when computing loop diagrams in perturbation theory. r(1/2) = V1f (4.4.1) where k is an even integer and m2 is a function of external momenta. Furthermore.5.7) follow from the definition. In the following. where it has simple poles. as usual. one computes the integral in a ddimensional spacetime.2.4 + E) (m2)(n4+E) 4(47r)2 r(n) x (gltV gPCT + gYp u" + s" gVCT) . except when z is 0 or a negative integer. I will show how to compute ultravioletdivergent loop integrals in dimensional regularization.8) 62 . is that of modifying the dimension of the integration space (spacetime in our case): the integral of 1/(k2 .5 Dimensional regularization A convenient way of regularizing divergent integrals.2 + E) (m2)(n2+E) (47r)2 r(n) 2(47r)2 (4.4.5. After Feynman reduction of the denominators and appropriate shifts in the loop variable. One finds (4. and then continues analitycally the result in the complex d plane. parameters.2E. d = 4 . For k = 0. while it would be convergent if spacetime dimensions are lowered to 3. The Euler r function is defined by (4.2) ltv 9 (_1)nl i(47rY r(n . it can be shown that r(z) is analytic in the whole compex plane z.5. and Feynman i(47rY r(n .4 we find (_1)n masses.5) where we have set.3 + E) (m2)(n3+E) r(n) (_1)n i(47rY r(n . Dimensional regularization is particularly useful because it preserves Lorentz invariance and gauge invariance of the theory. with d chosen in such a way that the integral converges. More generally.4) (4.3) (4. loop integrals can be reduced to the form (4.6) The properties I'(c + 1) = zr(z). I'(I) = 1.5.
12) 63 . Figure 4.if .+__.. 4. (4. with E = Jif + m2• of equal to zero..m2 + iTJ)n 1 = O.. = 0.1: Integration the Feynman in the complex qo plane..5.. Equations (4. We have therefore j+oo 00 dqo (q5 . (4. By virtue of the analiticity properties of the integrand in the complex qo plane. the qo integral along the closed path C shown in fig..5. we find (4. We now compute explicitly the integral in eq. + . Crosses indicate the singularities integrands at qo = ±(E ..iry).5.where d 'ljJ(s) = ds log I'(s] (4._ E+i"l x ..1 is ..5.5.5.m2 + iTJ)n 1 + J+ioo t': dqo (q5 ..5.5..11) With the variable change qo = iq4 in the second term of eq.10) 1 1 'ljJ(1) = .4..if . .5772 ...11)." 2 n (4.9) and 'ljJ(n + 1) = 1 + .4) (and similar formulae with higher powers of q in the numerator) can be obtained by shifting q + q + k and taking derivatives with respect to k at k = O.2). (4.3.+ .
which can therefore be integrated directly.::::m (47r)d/2 r( d/2) 1 1 (2) n+d/2 10 r(n .5.5.s.5.5.Notice that the +iTJ prescription is now immaterial.d/2) ( 2)n+d/2 r(n) m r 1 dx Xnd/21 (1 _ x )d/21 (4. we finally obtain (4. when one computes onshell amplitudes in a massless theory. is a vector in a 4dimensional Euclidean space.15) Thus.5. r(a+b) (4. 64 . We have therefore since the integration is performed along the (4. for example. 10 r 1 dx xa1(1 _ x)bl = r(a)r(b). On the other hand.18) which gives . Using this result.14) where we have used polar coordinates gaussian integration formula gives and the definition of r(z).16) For d = 2.3 the familiar results J d02 = 27r.2E. the usual (4.5. (4.20) By replacing d = 4 . We have (4. This happens.5.21 ) which is the announced result. imaginary axis. The integral over the ddimensional solid angle can be obtained in the following way.5.19) (47r)d/2 where we have used .13) where q in the r. J d03 = 47r are recovered.h. We first observe that the integrand does not depend on angular variables. Notice in particular that the integral vanishes when m2 = O. we have (4.17) The integral can be performed with the change of integration variable (4.5.
28) Repeating the same procedure for n = 3 one gets (4. for n=2. (4.29) to hold for any value of d.5. however.5.22). 15 (or equivalently the antisymmetric tensor EttVPCT)is a quantity whose definition is strictly connected to the fact that spacetime is fourdimensional. this requires 2n . and at each step a trace We denote by C2n1 the sum of such terms. 65 .5.22) hold in d dimensions as well. and we get (4. In fact.27) Using eq.29) For d = 4. ·'tt2n· (4. It is tempting to define 15 simply by requiring that its fourdimensional properties (4. To see this important fact explicitly.22).24) We can use the anticommutation rules {. (4.5. In particular.26). one would conclude that there is no axial current anomaly! We conclude that the definition of 15 cannot be based on eq. if we require eq. since it does not give the correct answer when d tends to 4.25) (4. At the end of the property of the trace and eq.26) Tr = and. by the way).5.5.5. (4.5.30) which is manifestly an inconsistent result. Tr 15 = 0 151tt21tt4 (4. procedure.23) for example.5. then we are forced to conclude that (4. the product.The use of dimensional regularization poses some special problems in calculations where the 15 matrix is involved.5. (4.2 I matrices appears.5.29) is satisfied for any value of Tr 151ttlltt21tt31tt4' which in fact is nonzero (and proportional to the axial current anomalous term. (4. (4..tt' IV} = 29ttv to bring. eq.1 steps.25) gives gttltt2 + C2n1 C2n1 = o. leads to inconsistent results.5. (4. (4.5.5.27) implies (4. and a definition in d dimensions requires special care. the trace can brought to T =T or For n = 1 eq.5. It is easy to prove that this assumption. eq.5. together with the circular property of the trace operator. using the circularity its original form. gttltt2 Tr 151tt31tt4  gttltt3 Tr + gttltt4 151tt21tt3 o. Ittl at the right of with 2n . consider the trace of 15 times an even number of I matrices: T = Tr 151ttl .
it can be obtained by the same way of reasoning that leads to eqs.5.14) we have (4.5. regardless of the value of d (this result is nontrivial.5.27).32) mixed components obviously vanish.j takes the ordinary values 0. (3.5.The correct way to define /5 in dimensional regularization is the following. which must therefore be restored by means of finite renormalization. Tr /5/I"/V =0 (4.39) which is the correct fourdimensional result.34) It is easy to check that the definition (4.1. I will present the computation in the massless case.33) Then. We decompose all/ matrices into a fourdimensional and an extradimensional component: (4. I will show that the computation of the axial current anomaly.38) vanishes if at least one of the indices has a value in the extra dimensions.5. one sees immediately that the quantity (4.34) requires special attention.2. Furthermore.5. Prove it as an exercise). We have therefore (4. performed in sect. The use of the definition (4.3 and 11" vanishes in the ordinary dimensions. that is /5 = '[/0/2/2/3·  (4. (4. The anticommutation relations become (4.35) or.3 in the PauliVillars regularization scheme.40) 66 .26) and (4. can also be performed in dimensional regularization.31 ) where "II" is nonzero only when j. we simply define /5 as in four dimensions. In the following.37) can be shown to hold. the matrix tensor gl"V has a fourdimensional and an extradimensional part. (4. the extension to massive fermions is straightforward. (4.5.5.3. I will not discuss this point in detail here.34) implies (4.5.5.5. because it introduces an explicit violation of chiral invariance. 3. in a more compact form.5. s'" = gl"V + fjl"V. Correspondingly. From eq.5.5.36) The identities Tr /5 = 0.
5.43) where 1= (4.50) 67 .5.5.tt~1~2 .46) exploiting symmetry under kl +t k2 and v +t CY.( 1+ d.2 2klk2 J) .42) involve the integral (4.tt~1~2. The numerator of the integrand contains terms which are linear. quadratic or cubic in the loop momentum k.2kvkO<Tr 1510<.k2)2(k + k1)2· (4.tt~(~1 + ~2) = 2k2Tr 15IV. (4. using the identities klk = ((k + kl? . (4.k2)/2.45) which can be written in the form (4.46) gives I~ I = dA + 2klk2B (klk2?B. where J= f k2 (27r)d (k .44) The second and third terms in eq.where the integral is made convergent by dimensional regularization. we observe that eq.5.5. The quadratic term requires more work.2kO<kttTr 15Iv'0<~1~2.49) Solving the system (4.47) one can show that (4. (4.5.48) k2k = (k2 .5. krk~Ivo< = klk2A+ Now. The first term contributes to the final result with (4.42) f (27r)d ddk 1 (k .5. (4. It is clear from eq. (4.5.5.5.5. one gets 1 A=.42) that only the term Agvo< contributes to the result. We have Tr 15~IV~. In order to compute A. The linear term is convergent.(k .47).5.5. and it gives a vanishing contribution: (4.k2?)/2.k2)2(k ddk + k1)2· (4.41 ) because k~ = k~ = o.tt~1(~1 + ~2) + Tr 15~2IV~.
5. It is now easy to compute (d4)I and (d4)J for d = 4 with the help of the formulae obtained earlier in this Appendix. Y"l Y"2.4 factor in front of the divergent integrals.33).Finally. 4.3. we come to the cubic term: since the anticommutator therefore compute term gives zero contribution because of antisymmetry. (3.5.4 ( I + 2klk2 d_ 2 J) Tr /5/ v / /LlJ.4 d .4.5. 68 . and there is a d . D(ka ka) 2 We must () t= = f a k (27r)d (k _ k2)2(k ddk + k1)2 = 1 . we finally obtain (k 1 + k)2 pT /LVP _.52 and taking the product Iak? one easily obtains (4. The final result is ultravioletfinite: indeed. and recover the result of eq. in dimensional regularization at one loop ultraviolet divergences manifest themselves as simple poles in d . u (4.53) Collecting all our results.54) where a factor of 2 has been inserted to take into account the contribution of Tlj'vP.