oy AC ty Bbs Gauge theory of = ‘°*" elementary particle physics Problems and solutions TA-PEL CHENG University of Missouri ~ St. Louis 7 2 8s LING-FONG LI - : Camegie Mellon University. wee Neg! L37ERY rot CLARENDON PRESS * OXFORDi OXFORD LURIVERSTTY PRESS (reat Caren Stet, Oxford OK2 6DP Orford Univenity Presi a departinet of the Urlverity of Oxford. ln rthers the Unlveruity's objective of excellence in resarch, x holarsp, ‘nd education by publlalng weotwide bn Oxford New York Athen Auckland Bogeth Buenox Aires Calcutta (Cape Towe Chcoral Dares Saluum Dette Horence Hong Kong fcanbul ‘Karacht Kuala umpur Madnd Mefboure MealcaClty Muriel NNalrobi Part StoPrulo Singspore Taipel Tokyo Toeonto Warnaw ith asociatd companies In Hern Ibadan (Oxford ins regserod tae snark of Oxford University Prest inthe UK and in certain other countior Published inthe United States by Oxford University Press ie, New York © Tea Chang und Ling- bong La, 2000 ‘The moral ght ofthe author have been asserted 1 atshase right Oxford Uneveaty Press (maker) {ona pubtaded 2000 All ight reserved. No part f his publication may be reproduced, stored In reclevalnywer, or rnamited, In un form of by any means, ‘without tbe prior permission m wring of Oxford Univeraty Press, ‘ax capronaly pected hy ln. oF unde terms agsoed withthe wppropeiate ‘eprographice rights organration, Paquinen concerning reproduction culise the scope ofthe above should be we tothe Rights Deparment, (Oxford Univenity Pent af the address nbove ‘You mont not clecul te thie bock In any utr nding ox cover ‘id you runt linpose th xame condition on uny soquier ‘A catalogue record for this book tx avaaable foe the Bela teary | brary of Congrens Cataloging ln Publication Data (aa eallble) isan 0 19 ws0021 Typeset by [Nevgentimagsytens (PL, In Grea iain Bkidles Lad, ford King yrPreface Students of particle physics often find it difficult to locate resources to learn calculational techniques, Intermediate steps are not usually given in the research literature. To a certain extent, this is also the case even in some of the textbooks, In this hook of worked problems we have made an effort to pruvide enough details so that a student storting in the field will understand the solution in euch case, Our hope is that with this step-by-step guidunee, students (after first attempting the solution themselves) can develop their skill, and confidence in their ability, to work out particle theory problems. This collection of problems has evolved from the supplemental material devel- oped for # graduate course that one of us (L.F.L.) has taught over the years it Carnegie Mellon University, and is meant to be a companion volume (0 Our text= book Gauge Theory of Elementary Particle Physics (referred to ax CL throughout this book) rather than a complete assemblage of gauge theory problems, Neverthe less, it has a self-contained format so thut even a reader not familiar with CL ean use it effectively, All the problems (usually with several purts) huve been given a descriptive title. By simply inspecting the tuble of contents readers should be able to pick out the areas they wish to tackle, Several new subjects have entered in the field in the fifteen years since the original writing of CL, Although we have not revised the book to incorporate them because we would not be able to do them justice, we hope this set of problem! solution presentations is the first step towards remedying the situation, We have incorporated a number of new topics and developed further those that were only introduced briefly in the original text, Listed below are some of these areas: ‘¢ Relations among different renormalization schemes: Further applications of the path-integral formalism General relativity us a gauge theory Superconductivity as a Higgs phenomenon Non-linear sigma mode! and chiral symmetry Path integral derivation of the axial anomaly Infrared and collinear divergence in QCD Further examples of the parton model phenomenology QCD and A/ m § rule in the non-teptonic weak decays More on gauge theories of lepton number violation Group theory of grand unification Further examples of solitons Many people have helped us in preparing thix book. Our thanks go particulirly toall the students who have taken the course and have worked through a good part of these problems, One of us (T.PC.) also wishes to acknowledge the enjoyablevi Preface hospiuulity of the Santa Cruz. Institute of Particle Physics when finishing up this Project . The original literature has only been referenced casually, and we apologize to the wuthors whore work we muy have neglected to cite, This book and CL share a page on the Work! Wide Web at the URI. http: //www.umsl.edu/~tpcheng/gaugebooks -html. Misprints or other corrections brought to our attention will be posted on this page. We would be grateful for any comments about these books. St. Louis Pittsburgh January 1999ntents Fleld quantization Simple exercises in Ag* theory Auxiliary field iscomnected diagrams Simple external field problem Path integral for a free particle Path integral for a gencral quadratic action Spreading of a wave packet Path integral for a harmonie oscillator Path integral for a partition function Partition function for an SHO system Non-standard path-integeal representation ‘Weyl ordering of operators Generating functional for a scular field Poles in Green’s function Renormalization ‘Counterterms in 4g theory and in QED Divergences in non-linear chiral theory Divergences in lower-dimensional field theories n-Dimensional ‘spherical’ coordinates Some integrals in dimensional regularization ‘Vacuum polarization and subtraction schemes Renormalization of 4g" theory in n dimensions Renormalization of composite operators Cutkosky rules Renormalization group Homogeneous renormalization-group equation Renormalization constants B-function for QED Behaviour of near a simple fixed point Running coupling neur a general tixed point ‘One-loop renormalization-group equation in massless A¢* theory B-function for the Yukawa coupling Solving the renormutization-group equation by Coleman's method Anomalous dimensions for composite oper tlors 7 x» 4 43 46 49 53 an: 539 63 67 oo 0 7 an 15vili Group theory and the quark model Unitary and bermitian matrices SU(n) mattices Reality of SUQ) representations An identity for unitary matrices An identity for SU(2) matrices SU(3) algebra in terms of quark fields ‘Combining two spin-5 states The SU(2) adjoint representation Couplings of SU(2) vector representations Isospin breaking effects Spin wave function of three quarks Permutation symmetry in the spin-isospin space ‘Combining two fundamental representations ‘SU(3) invariant octet baryon—meson couplings Isospin wave functions of two pions Isospins in non-leptonie weak processes Chiral symmetry Another derivation of Noether's current Lagrangian with second derivativ Conservation laws in a non-relativistic theory Symmetries of the linear a-model Spontaneous symmetry breaking in the o-model PCAC in the o-model Non-linear ¢-model | Nor near o-model I Non-linear o-model 111 SSB by two scalurs in the vector representation Renormalization and symmetry Path-integral derivation of axial anomaly Axial anomaly and n> yy Soft symmetry breaking and renormalizability Calculation of the one-loop effective potential The Parton model and scaling ‘The Gottfried sum rule Calculation of OPE Wilson coeflicients Sale e> > hadrons) and short-distance physics OPF of two charged weak currents The total decay rate of the W-boson Gauge symmetries ‘The gauge ficld in tensor notati Gauge field and geometry General relativity ax 2 gauge theory Contents 8 ” 9 81 82 83 85 87 89 90 93 96 97 100 105 107 no mw 3 WS 122 123, 126 128 130 133 136 140 142 143, 146, 147 151 155 156, 158 tol 163Contents: O(n) gauge theory Broken generators and Goldstone bosons ‘Symmetry breaking by an adjoint scalar ‘Symmetry breaking and the coset space Scalar potential and first-order phase transition Suiperconductivity as a Higgs phenomenon Quantum gauge theorles Propagator in the covariunt Ry gauge The propagator for a massive vector ficld ‘Gauge boson propagator in the axial gauge ‘Guuge boson propagator in the Coulomb gauge ‘Gauge invariance of a scattering amplitude ‘Ward identities in QED Nilpotent BRST charges BRST charges and physical states Quantum chromodynamics Colour factors in QCD loops Running gauge coupling in two-loop Cross-section for three-jet events Operator-product expansion of two currents Calculating Wilson coefficients Electroweak theory Chiral spinors and helicity states, The polarization vector for a fermion ‘The pion decay rate und fy Uniqueness of the standard model sealar potential Flectromagnetic and gauge couplings Fermnion mass-matrix diagonalization An exumple of calculable mixing angles Conservation of the B= L quantum number Electroweak phenomenology Atomic parity violation Polarization asymmetry of 7 > ff Simple z-lepton decays Electron neutrino seatterings CP properties of kaon non-leptonic decays 2» Hi is forbidden Al = § enhancement by short-distance QCD Scalar interactions and the equivalence theorem ‘TWo-body decays of a heavy Higgs boson 165 167 169 m 172 173 175 1%6 77 178 180 180 184 186 188 191 193 198 201 205 206 208 212 213 214 215 216 218 221 222 223 225 226 227 230 234Topics in flavourdynamics Anomaly-free condition in a technicolour theory Pxeudo-Goldstone bosons in a technicolour niodel Properties of Majorana fermions > ey and heavy neutrinos Leptonie mixings in a vector-like theory Muonium-antimuonium transition Grand unification Content of SU(S) representations Iliggs potential for SU(S) adjoint scalars Massive gauge bosons in SU(S) Baryon number non-conserving operators SO(n) group algebra Spinor representations of SO(n) Relation between SO(2n) and SU(n) groups Construction of SO(2n) spinors Magnetic monopoles ‘The Sine-Gordon equation Plunur vortex field Stability of soliton Monopole and angular momenturn Instantons ‘The saddlc-point method An application of the saddle-point method, A Euclidean double-well problem neces Contents 238 239 239 244 250 252 255 256 258 260 260 263 267 275 280 282 289 295 301 303Field quantization imple exercises in X¢4 theory In A¢* theory, the interuetion is given by t= Xo, a) (a) Show that, to the lowest order in A, the differential cross-section for two- particle elastic scattering in the centre-of-mass frame is given by (1.2) wheres m (prt p2)*, with py and p2 being the momenta of the incoming particles. (b) Use Wick’s theorem to show that the graphs in Fig, 1.1 have the syinmetry factors os given, Also, check that these results agree with a compact expression for the symmetry factor S=x [T] 2% meen * (13) where g is the nuniber of possible permutations of vertices which leave unchanged the diagram with fixed external lines, a, is the number of vertex pairs connected by 1 identical lines, and is the number of lines connecting a vertex with itself, (e) Show that the two-point Green's function satisties the relation (By + 2 MOITH(XP(YNI0) Sear tngonin - is -y). Also verify this relation diagrammatically to first order i aS 0 D sax Pra. 1.1. Symmetry factors. AW2 Fleld juanitzation 1.1 (d) A Green's function involving the composite operator Q(x) = 467(x) is defined as GM on. = OIT(QWP Er) «+ GU%n))0)- «4y Write down, to the first order in A, the various contributions to G2? (xs x14.X2). Solution to Problem 1.1 (a) The tree diagram for a two-particle elastic scattering is shown in Fig. 1.2. Thus (othis order the scattering amplitude is simply 7 = —i2 giving rise to adifferentiul cross-section; (see CL-Appendix A for rules of cross-section calculation): es ee 3) de = yl Er 2! apa Cae 2B x(2n)'8\(py + pr — pr- pat. (5) The last $ fuctor is inserted to account for the presence of two identical particles in the tinal state, ‘We then have the phase space factor of apy dps (2x )82Ey (2x 2E, where 6 = pr + pr. In the centre-of-muss frame, the four momenta cun be parametrized as p, = (E,p), p= (E, ~p). py = (F', p’), and py = (E’, -p'). p= famsi—m—po 6 After performing the dp, integration, the phase factor becomes dp . 306-26 P. p= fan 308 - 2695 EGE’ = [em-s2e- 262 f (2m)*5(2E — 26) = IPI =a (7) and thus the differential cross-section do) 1 pl (ay dQ Wi = v2] 4B? 3207 E 2° a ay tk a MeLL Simple ex reixes in 4* theary 3 Alter substituting the flux of |v; — val = |(Pi/E1) — (P2/E2)| = 2ipl/E and the invariant variable s = (p, + p2)? = 4E? into the above expression we obtain jo Trent” (b) (1) The diagram in Fig, 1.3 corresponds to u second-order term in the pertur- bution expansion a9) a x (#) f Ay dy OT OOO OE ONEONEOD: (ODO OOOO). (10) The amplitude like the one above but with the interchange ys ++ yz has the same contribution. This doubling cancels the first factor of 4 in the above expression, which comes from the Taylor expansion, Wiek's expansion leads to the following contractions, There are four ways to contract #(x1) with any one of the #(y1)8and similarly four ways to contract (x2) with any one of the o(y2)ss then there ure 31 ways to contract the renuining pairs of (yi) tnd O(y2). The (inverse) symmetry fuctor is aay ‘This checks with the result obtained by using eqn (1.3) directly, because g = 1, ay = Lund 6 = 0. (Hh) The diagram in Fig. 1.4 is first order in the coupling 3 fey {O17 19(4199(42):(18 YH N1HO):I10). (1.12) There ure four ways to contract #(.x,) with any one of the #(y)s and three ways to contract @(x2) with any one of the remaining three (y)s. Hence a3 . (1.13) This checks with the result obtained from eqn (1.3), since g = 1. of = 0, and fol. Pa. 14,4 Field quantization 1.1 (iii) The dingrant in Fig. 1.5, like the one in Fig. 1.2, corresponds to the second- order term as given in (i). The mohiplicity is determined by noting that there are four ways to contract $61) with any one of the (y1)s and three Ways to contract (x2) with any one of the remaining three f(y1)8. And there are (3) = 4+ 3 ways to contract the remaining (yr) pair to all the possible pairs out of the four $(¥2)s. a 1 1 s (5 )ae te 3s (1.14) Fquation (1.3) ulso yields $= 4 because in this case g = 1.a2 = 1.and B = 1. (lv) Figure |.6(a) is a fourth-order diagram. There are $ such diagrams corres- ponding to 4! ways to permute the yr.2..4 positions foru fixed 1.2.3.4, ard the wo categories of diagrams a iMustrated in Fig. 1.6(b) are actually identical, Thus the ‘Taylor series factor of jy is only partially compensated. For brevity, for the remaining part of the amplitude we will only display the position fuetors of the tields XURKKE MYLYIT YW WMS Yayayaya (sy n 4 4% Po LeLL Simple xercises in g* theory 5 and examine its combinatorics. There ure 4 + 3 ways to contract (xi) und p(x2) with the four (1), and the same number of ways between (x1) and (x4) and the $()2)s. For the remaining two $(y1)s to contract into the respective four O(y)s and G(y4)s there are 2-44 ways, Similarly, for the remaining two $(y1)s to contract into the respective remaining three o(y3)s and @(y4)s there are 2-3-3 willy, there are two ways for the remaining two $()3)8 and $(y4)s to contract into each other. oa 1 (3-4) a (4-3)? (21404) (2-33)-2 _ 14:3 2 “27 (a This again checks with eyn (1.3), since g = 2.c2 = 1, und 8 = 0. (6) First we show that the differentiation of the two-point function with respect to x yields (1.16) HOT POSEN) = OIF" GHP OIO) F#UOIG(X). HCVIMO}S(X0 = Yo) 1-17) where the equal-time commutator actually vanishes. Differentiating for the second time we have OOTP (0G (VIO) = (OLT(AP(x)9(y))10) +4006 (x). P(yIIIO)8(Xo — yo). (1.18) From the equation of motion Og(x) = —p2p — (4/31)¢* and the canonical commutation relation [dy(x), ¢(y)18(t0 — yo) = —184(e — y), we then obtain the result stated in the problem. a Ce + POT OONIO) = — FOF AIS(N)IO) — 184(x — 9). (1.19) ‘To verify this relation diagrammatically we note that the first order in 4, diagrams for the (wo-point function ure given in Fig. 1.7(a). ‘The Feynman diagrams lead us to the relation CIT HCOPL)IO} = AKG = y) + (¥) x ftir —omiarte— yar (20 Using the relation (O, + u?)A p(x — y) = -54(x — y), we obtain the left-hand side of eqn (1.19) to first order: i, wists - 9+ (FP) f atetote-omiare- pao - a hare = VAL(O) = 184(x — ¥). a26 Field quantization 1.2 x wor ky iy: @ o Pe. 17. Tae) ) O) Fo. 18. ‘Writing out the Green's function on the right-hand si a we have A IT (D(x) (XIP(XIG(Y))IO) = lhe — yi Ar). (1.22) Equations (1.24) and (1.22) clearly show that the relation (1.19) is satisfied. The Feynman diagram for egn (1.22) is shown in Fig. 1.7(b). (d) There are thrce first-order diagrams for the two-point function Gs 1s ta) = (O17 367 ODH (414 (42))10). (1.23) We shull explicitly work out the cuse of diagram (a) in Fig. 1.8. 7 f d*yg"yy10) (017 $870) (x 16042) wa ia =(F) [ atotiarce -yracen— stare yP. 0.25 ‘The symmetry fuctor of S = 2.can be understood by noting that there are 4-3 ways to contract between # (x1)¢ (x2) and two ds in ¢*(y), and 2 ways to contract $2(x) with the remaining two #(y)s. Thus the 4! factor in the coupling is cancelled, and we are left with the $ factor from the composite operator. The diugrams in Fig. 1.8(b) can be worked out in the sume way. Their symmetry factors are also 2. Auxiliary field ‘The Lagrangian density for a set of real scalar field @*,a = 1, 2,. ! ? yon — he = 3a,0 06") - Fare" — Fore? (125) int ampu- (u) Work out the basie vertices in thi theory by calculating the four- tated Green's function to the first order in 2.1.2 Auxiliary field 7 (b) Consider the Lagrangian density 5 e849") — where @(x) is another scalar field. () Show that if we eliminate o (x) by using the equation of motion, we end up with the Lagrangian in en (1.25). (i) ff we do not eliminate o (x), and take the propagator for a(x) in the momen- tum space to be —i2 (which can be justified by adding a term (€/2)(2,0)(8%c) d then the limit of ¢ —» © after the propagator has heen worked out), show that £ gives the same basic vertex for g(x) as that given in part (a). Solution to Problem 1.2 (a) To the first order in 2, the four-point Green's function with the four external lines carrying the itternal indices a, b, c,d is given as org*e'go! (=) o'e'o!o!i0) (1.27) where we have grouped the four fields in the interaction term into two pairs labelled i und j, respectively—repeated indices are always summed over. "As displayed in Fig. 1.9, there are two ways ¢'s can be contracted with ¢%9", and two ways between ¢/s and ¢ 94; these four ways are to be multiplied by 2 corresponding to the interchange i «+ j. Thus the factor of 8 is cancelled und the vertex is given by -iAd8", There are of course other ways we can pair off the four external lines. Removing the propagators for the external lines, we have the basic vertex for this theory: win (ata + Ber8™ + gts), (1.28) (b) (i) Since the £’ does not contain the ,¢ field, the equation of motion for the © field 3L'/aa = 0 is simply a constraint equation: o/A = 44°¢", Substituting this condition into the £' Lagrangian density, the dependent part becomes as R and thus £' = £. @ory (1.29) 24% we Ngee? — digg? a oo" grey qeey8 13 - Ge) FLL 1.10. Feynman rules forthe Lagrangian C’. a b a a 72 ‘ ; 4 1. 2. \. ¢ 4 ¢ ‘4 @) (by (ce) fo. Lt (i) The Feyamun rule from the £’ theory is shown in Fig. 1.10. From this we can construct the Feynman diagrams for the four-point function in Fig. F.1 Diugram (a) yields (-/5")(+i)(-i8%) m= —i8%5°4, Similarly, diagrams (b) and (c) give 16" 8“ and —i86", respectively. Remurk, Very often this kind of auxiliary field is introduced (o make the calcu- lation more tractable. For the case here, the use of the @-field makes the flow of the internal symmetry indices easier to monitor. Isconnected diagrams Consider the unperturbed und perturbative parts of the scalar field theory elegy Ma fm e Com 50-Fe, C=- Se, (1.30) In the perturbation theory, the two-point Green's function is given by G™ xr, x2) = (OTC 64190(42))10) (O17 ota bo(x2) expl—i fH’ dx 10) (OI Cexpl=1 f Hey AxDIOD Use Wick's theorem to demonstrate explicitly that the respective disconnected graphs in the numerutor and denominator cancel a314 Simple sternal field problem 9 Solution to Problem 1.3 The two-point Green's Function (O17 (ox boCx2) expl—=i fH’) dx})10) NV ue GM ex..m2) = oir explai f HG)axDI (1.32) has the following Wick’s (diugrammatic) expansions, shown in Fig. 1,12, for the denominator D and the numerator NV respectively, where the dot represents the *vertex* of Ly = —(n?/2):82(x):. This is equivalent to the expansion ax shown in Fig, 1.13. We see thut the disconnected contribution has been cancelled. imple external field problem Suppose the Lagrangian for a sculur field is given by g? = J()6() (1.33) I 1 Lazo? - where J (x) is a real ¢-number function.” Fi td.quaniteaion 1.4 (a) Calculate (0[6(x)10) and the two-point function (UIT (g(x) (O))|0) exactly. (b) Treat the term J (x) (x) as a perturbation und calculate (0|¢(¢)/0) and (017 (6 (2) 0))10) to the lowest order in J(x). Solution to Problem 1.4 (a) The Lagrangian yields the equation of motion (0+ WG (x) m= I(x). (1.34) If we define the usual Green's (propagator) function, (+m Arte - y) = 8x y), {he field operator can then be written as 9(x) = a(x) + (x), where Gn(x) is a ¢-number function: ducers f tty arte — naen. (1.36) G(x) satisfies the homogencous Klein-Gordon equation, (0 + 2)6(x) = 0, and canbe expanded in terms of the usual creation and annihilation operators, satisfying, the commutation relation [a(k), a"(k’)] = Bk’): : ak de) -[—*a fackye** + atte), (137) [amen] [ ] Because $(x)10) = (016(x) = O the vacuum expectation value of the unshifted field operator is non-vanishing: (1610) = box) = f dy Ar(x— YS(Vs (1.38) tnd the two-point Green's Function is also shifted as {O17 6G (ODIO) = (O17 GoxIG0(O)10} + OITSEDBCODIO) = dulr)do(0) +A r(x). (1.39) (b) The ‘interuction vertex’ in the Feynman's diagrams for this theory is given in Fig. 1.14(a). (1) The perturbative expansion for the vacuum expectation value cun be repre- sented by a diagram similar to Fig. #.14(a). recom = Corre OS x fen JOVEN dnd Onn) = -ifey {OT Ortx)br O1OI(Y = f trance = YY) = bol).LS Push integral for a free particle " Jo Jy) JOD * ¥ x trees | 0 ® ) Ta. 1.14, (lI) The perturbative expansion for the two-point Green's Function is given by Fig. 1.14(b) O1TP(xIG (OVO) = {O17 G1(x)61(0) : &S [end'nsonsoveineonI0 an = SP fan Ay FA rx — yde(—w OO” +019 wl = bax )60(0). (1.40) ath integral for a free particle Show that the transition amplitude for a free particle (mass m) moving in one dimension has the expression 1/2 @ ] evo| " You should check that this result can be obtained by starting either from the Hatniltonian or the path integral (Lagrangian) representations of the transition amplitude: rl) leon (aly (g'lexpl-fH (' — ollq) : a2 W fldghexpi ff ar*t] > ¢t'lg.) = | where H = p?/2m and L = (0n/2)4? and the integration measure in the path integral representation is given by Moda = eg (si) Ta aay with (1 — 1) being divided into a equal segments of A: 1,1. 12, font Mae having the corresponding positions q. qi. gas +++ @a- 10 9" ME Gn-12 Field quamization 1.5 Solution to Problem 15 (a) The Hamiltonian method a’. tla) = (g'lexpl-iH’ Nig) a -ip =’ jean Su -o] 9). Inserting a complete set of momentum states: -o lp)tpla) ep | Eu -nsipw’-a i om id's eae 4 which cun be integrated by using the Gaussiun integral formula: iad e ax? = fr exp(2 [trex ax? + bx) (Ee0(Z) In our case, we have a= (i/2m)r' = 1) and b = i(g" —_g). Thus, ee No. = m im Gg" = a) wats [so] oo( ror |. (b) The puth integral method ‘The action can be written in terms of the spuce-time intervals ox fie [ee aay s=[ La f Sear sr(oe 4 m = ala -a tna tet Get 9). Using this and the given integration measure, the transition an expressed ax Gta. iy” [Elen on [F(a -a0? + Gig tet Grit —9] |. (1.44) 1.48) (1.46) (1.47) (1.48) ide can be 49) ‘The successive integrals can be calculated by using the formulue for Gaussian Integrals of the form re 70 cil ab ft. deexp[a(x — xy)? +.5(¢ =?) = (S] on [5 - x]1.6. Purh imegrol for a general quadratic action B 0 that one hus fener im ig-arttwn olf 24 i 2 im (g - a3)? - [24.3] oo al meee: [ene { a aa +a -a]] 2d 27'? Fim y — gs? m 3 2° 3 oo fenoo{ gra) © ail [ms 3} [2 | =/(= exp| = m4 20 tnd s0 on, In this way, one obtains n\n (2nia\ ODA ay FA) im a xon[ He -0] : m via im(q’ — q)* = (ma) 9 [ Se | m__\ Trimet — 4? 2 no BLL ak be 50 (azn) on 201) ] ew where one hus used nm (1/1), This result agrees with that obtained in (i) by Using the Hamiltonian ws the generator of time evolution. th integral for a general quadratic action ‘We will study the case of the action containing at most quadratie terms Sil = fu lating? + bing + ctngg + alg +e(Dg? + f«}. LSD) (a) Show dat Maret plant) me Ft exp [iSelgp. ty: diet] (1.82) where S(gy. ty: gis h) isthe action forthe classical trajectory, and F(t. 1,).being independent of q and q , cun be written ax tunwen fo ‘anorexe [f° dt fait toni ter} s3) toa 5 Namely, we have the boundary condition (1) = n(’) = 0. Thus (1) can be thought of ux the difference between » given q(t) and its classical trajectory.4 Ki ld quantization 16 (b) Show that the prefactor F(t, 4) can be expressed in the compact form of Fz.) = N’fldet Ay", (154) with the differential operator A = -a(d?/dr?) + c(d/dr) + and N’ being a constant, Suggestion. Expand n(r) as a series in terms of some orthonormal baxis functions Ault) (With n= 1,230.98 =D axl (1.55) where fil xu )%m(1) = Bun ANd Xnlli) = Xn(t7) = 0. The integration measure N{dn(t)| = NTT, dn(tq) cun be taken as NT], dcq. Thus we can obtain an alternative definition of the path integral as lan telat = W fT] tevexpista (1.56) Solution to Problem 1.6 (a) The path integeil representation of the transition amplitude has the form Mee ty) . . Captylqetd = wf Uiglexp? f dritia.dent. (sn had Let q (1) be the solution to the equation of motion 85 dat) ob ywto £(%) a=? (1.58) with the boundury condition qe(ti) = qi and qe(ts) = gy. An arbitrary path g(t) can always be written as q(r) = qe(1) + n(r). Namely, n(t) is defined 10 be the deviation of q() from the classical trajectory with the boundary condition n(u) = n(tz) = 0. In terms of the unique classical trajectory and (2), we can express the transition amplitude as cay) (gy. telqien) = wf [dn(eexpliSige + nll. (1.59) qu? ‘The action S can be expanded in powers of n(1): Slge+n] = Sige]+Si +S, where 5) is linear in (2) and Sp is quadratic. Since the classical trajectory, according to the vartutional principle, corresponds to the path with respect to which the is surtionary, we have 5; = 0. Thus Slae-+ nl = Stq.l+ J " dt [ati yi? + clang + etn?) - (1.60)1.6 Path Integral for ag nerat quadratic action 8 Sgel is independent of n(¢). Evidently S; is independent of q., hence also of gi and 74. (This is only true for a quadratic action.) One then has a (aystslaet) mexpuista nn [ dnl 4) exp ff’ dt [a4? + ong + of = Fey. wexpliSaptp awn) (61) which is just the claimed result, Remark. In many physical applications of the path integral formalism it is not necessary toknow the prefactor F (17, #;), which does not depend on the coordinates eqs (b) Start from the expression . FU, = © fudnorem{i f dt (oi? +eni-+er?]} A th la eid = fue mtrtexe |i f one | ann te7 +4 no} 62) where to reach the second line we huve performed an integration by parts, Now expand n(¢) in terms of a complete set of orthonormal functions: n(1) = La enXa() with the condition of x, (01) = Xn(t7) = 0. We then have 7 . Funan f T]devero |i f dr nordnio (1.63) n Ih where A is the differential operator given in the problem. For convenience, we can choose the orthonormal functions to be the eigenfunctions of A: Arott # ely = hata 1.64) Xn(t) aR on Xnlt) = kn Xnl. (le Then oo dentydntt) mY f dt cacaknxalt%m(t) = Yo Aka 1.65) f nyAnte) z/ Wextabm tain) = Dh (1.65) 4and the prefactor for the path integral becomes Foyt = NT] f devexn (Dau). (1.66) - 716 Held quantiz uion 17 For each term with n = / we can use the Gaussian integral f dx exp(-ax?) = (t/a)! o obtain the promixed result: a ry.m=NT](F) = N'det A“ (167) Wecan check this result by working out explicitly the simple cuse of a free particle S = fi dr(m/2)q?. Thus A = —(m/2)(d?/(d#?) and the eigenvalue equation has the form for a simple harmonic oscillator equation, —(m/2)(d2xq/d1?) = kn Xn which has the solution of x, = a, sinw,( — 1). with w, = (2k,/m)'?. The boundary condition «(tr = 4) = m7, with 1 being an integer, implies that the eigenvalues ky = (m/2)(nx/ty — 4)? Thus, the determinunt has the value of det A-Y? w= [], kz", Include the multiplicative factor from the Gaussian integral of (/2)(i)"7 and choose the normaliz wion factor N! = [2n/im(ty = Wy"? 50 that we obiain the expected value (as determined in Problem 1.1) for the prefuctor: Fopnen'T] (42)? (4 / rr )=[axtsal (168) Preading of a wave packet ‘The time-dependent Sehrdinger wave function ix defined by y(q.) = (q.tI¥). (a) Show that Wart feartrlaneantd da (1.69) (b) For the free particle, suppose y(q. 1 = 0) is a Gaussian wave packet: a _ vom (52) on[ 87 | (1.70) Show that it will ypread as time evolves: 1 ye [ q-a? > = | ——— —S>S Iq. al exp a) aay where 2 one (1+ 53) ny Remark. One may recall the physical interpretation for this spreading Gaussian wave packet. The initial Gaussian wave function can be thought of as a super- position of plane waves e'*. As they evolve with time, such plane waves ucquire a momentum and time-dependent phase e~?/2 which will makethe superposition go out of phase for 1% 0.L&_ Path imegral for a harmonic oscillator 7 Solution to Problem 1.7 (a) This connection between the initial and final wave functions by the transition amplitude can be obtained simply by inserting a complete set of states in the expression for the wave function, Way. ty) = gp tl) = feacartnaomanlv. (1.73) {h) Substituting into the above equation the expression for transition amplitude for the free particle case us derived in Problem 1.1 und the wavefunction y(q. 0). m ya im(q-q'P tig. 0) = (<— fe : (a.tlg.0) = (57) [3 ; 74) we end up with a Gaussian imegral of the form shown in Problem 1.5. my fimg = =a Je @ el wy. n= fa( = oo[@ (3) exp] = =(0 ( 1 \M0 aniotr ? g-a? Nani) \3rot) Lire amat P| 402 + i2r/m ‘The yy has a simpler expression; itis struightforward to show it checks with the renult given in the problem. (1.75) Path integral for a harmonic oscillator ‘The Lagrangian is given by (1.76) (a) Show thut the transition amplitude has the form: ma fo Gap telaie tm sar sineo(ty =) ronan amy (1) Show that for an initial wave packet of the Gaussian form NC ¥i(q.0) = (ME) exp [Sg - a9] (78) we have Weg. = (22) exp[-marg —aeovew]. 1.79) Namely. there is no spreading of the wave packet.8 Held quantization 1.8 (©) In gencral, the transition amplitude, as u Green's function, can be expressed in terms of the energy eigenfunctions as a's th) = So tng OG (gd et (1.80) where by (q) = (qln) and #|n) = E',|n). Show this and then work out the ground sie energy und wave function by wking the limit of = 0 and ¢’ -» —foo in the transition amplitude (q'.1'lq.t). Solution to Problem 1.8 (w) The action being a quadratic function, the transition amplitude, according to Problem 1.6, has the form of (aptelgit) = Fly ndexp[iScgy.tes qt). agi) ‘Thus we need first to calculate the classical action S,, then the prefactor F Given the Lagrangian we can immediately write down the equation of motion: @ + wg = 0. Its solution corresponds to the classical trajectory: g-(N. Asinat + B coset with its coefficients A und B to be fixed by the boundary conditions of q. (¢ = fj m0) = q, and ge (t = 17) = gy. We find B= g, and A= (47 — 4, cosets)/ sinwt . Thus, qs ay [ar sineotty — 1) +7 sinor] (1.82) with the velocity . » ql) = Tnaty [-g) cosas —2) +4 concer]. (1.83) The classical action is Sgp.tys q1.0) = sf dt [920 - 0g] dt (q} cos 2w(ty = 1). = Show a3 i" tg} con 2u(ty = 1), + qy OR et — 2qigy cosar(ty — 21)]. (1.84) Itin straightforward to do the r-integrals wo sin wt, sin 2 sin wt Seago te: 00) = it [ame +e ane ae] "Tne [a? + gjreosanty - 2q47]- (1.85)1.8 Path integral for a harmonic oscillator 9 ‘This is the expression for the classicul action that appears in the transition amplitude: Lapetylquetsd = Py ndexp [Seay rtp qt). (1.86) Now we are ready to calculate the prefuctor F(ty, t,). It ean be determined from the condition of y= f dataytraethatlanso (187) (aps tsla where we have inserted a complete set of state {Iq. 1)) for a fixed time ¢. Explicitly writing this out, we have Fy. exp [iSelgs.ty qe) = Pap neem f dg expliSaartsiao+ ISG. git) (1.88) The integral on the right-hand side has the form 00 myn RR dq exp(—Ag? + Bi ]e= = exo (+6) (1.89) . in ig expr iy 4 G) 4A where Fea coset =) _ esol 4) : (1.90) a Lsinolts—1) ” sina? —n) For our purpose of calculating the prefuctor, there ix noneed to work out the Bg-+C term as it only contributes to the exponent, which must match the iS-(g 7.1/3 qi.) factor on the left-hand side of the equation. Thus with the presumed cancellation (check this) of the resultant exponential factors on both sides of the equation, the only relevant Gaussian integral is [erent = (5)"=[ ‘This means that the prefactors have the following relation: 2 ] - G91) mw yt 2nt m= lsinwtt, — 2)? F001) [sine — 4)! 2F G1). (1.92) a ? Exim otty = 4)]!? Fey td From the above equation we can deduce the desired result: ‘mes 2 lets = ll : - Fyn20 Field quannzation 1.8 (b) From Problems 6 and 7, we have the relation veg.nm fddtqtl’ Owego 9) where : mo yh ; utly’.0) = [2] J’ e0[ 2, [@2-+ 9% cose — ae'| (1.98) and, as is given, ‘ men '/4 mo, 2 valg's = (=) exp Sua]. (1.96) Putting them together, oe ey" (a hay iin) J & nero | Mf? +4) coset = 244] - ye -o'I. (197) “The exponential integrand having a quadratic funetion of ¢’: (+ Aq? + Bq’ + C asits exponent mw _ im tar 2sinwt rf 2sinwr eg tiasinet) (1.98) nwt ymca cOsa og _ Md a Zain or 2 the integral is of the Gaussian type discussed in part (a) und yiclds the result of (n/A)'? exp((B?/4A) + C). We obtain the final wave function _mey'4, mo desinot _j.,]'" vig. = (2) Ley expt] where the exponent has real and imaginary pars (1.99) nwo! 2sinwr (q +iasinwt)? + =-Fa acoso? +i sino Qag +0? cose), (1.100) tn this way we obtain the wave function “ wD 2 (Mey! Senet Oe viene (2) exp[ Seq — acosany] xexpi ["F sinen(2ag +a? eoswn]. diol and an expression for |%e(q. 1? just as that quoted in the problem. There is no spread of the wave packet because the originut Gaussian wave function is an1.9. Pash integral for a partition function 2 eigenfunction of the SHO Hamiltonian, and the time evolution comes in only as an overall phase. (c) We will first express the trunsition amplitude in terms of the energy cigenfunctions: (q's t'lq.t) = q'lexplit’ — lg) = DT exptt He’ = Nllnynigy = Dok hOMg'inyenigh = bala oz q) ef? ‘Sctting ¢ = 0 for convenience, it is clear that in the limit of r' m= -i7 with T + o this sum is dominated by the ground state |0) contribution: fim ig’ ree This should be compared to the limiting expression for the transition amplitude obtained in purt (a) io mo (| ii pa ro) = ( J exp | x [ig + a? yeoset’ — 24'aH} . 2ai sina 2xin ear! (1.103) Noting that both cos ct’ and i sin wr! increase as fer” in this limit: _ mw ya Mh ae W.-i719.0) = (25) evo | Fle +g fh oT - 2g] ‘meni TM 2 4 42) | e*T?. =(*) on eee Compare the expressions in eqns (1.102) and (1.104) and we obtain the ground state energy und wave function as dug) — (2)"a0(-420). (1.108) th integral for a partition function Show that the partition function of canonical ensemble Z = Tr( ~F) with f = (kT)"" and 1 the Hami for a simple case of one degree of freedom system, can be written as a puth integral representation as ip 2m fda fracrero(s f an@d) (1,106) i where gy = q(t = 0) = q(—iB) and A(q. 4) is the Lagrangian in E twit lidvan time2 Hletd quantization 1.10 Solution to Problem 1.9 The trace in the partition function can be written as the stim of the matrix elements of the e~>" operator between the cigenstates of the g operitor, Z = Yael “Igo. (1.107) © Compare this with the path integral representation of the transition ainplitude, ef. in particular CL-eqn (1.47), Kags tyldreh) m tgs Iga) “ = fda frarrerp(: f dip - Hq. nl). (1.108) A ihe partition function in eqn (1.107) can be viewed as the special case of the path integral representation in eqn (1. 108) with the restriction y-uenip oor y= 0 and ty = if (1.109) qi = q> Namely, the path q(1) is and the initial and fi periodical I position identified q dy) = GH) = do CTO) For convenience, we can introduce the new variable = it. The integral in eqn (1,108) becomes ip B dq if drlpe- nao f de [inst - Ha. p)] ay A 0 ae and therefore * de Zz = fitai fapie if de [ies -Hw. ol}. aun) For most cases, H = (p?/2m) + V(q), the momentum integral is Gaussian fowl bene] ovo| f"ae{-3 (44) -»}} (3) ‘And we have the partition funetion zan fuse ols (4) +] aa where the combination —[(m/2)(dq/dr)? + V| is just the usual Euclidean Lagrangian A(q, 4). The constant NV is independent of temperature (hence has no physical significance). The integration is over all the periodic paths with the boundary path go = q(0) = (A). From this problem we sce that the partition function ean be obtained from the usual path integral method through the steps (i) set 4, = 0, and ty = —iB, and a(t) = q(t) = qo: and (ii) integrate over gy.1.10 Partition fiction for an SHO system B Partition function for an SHO system The partition function for the ease of the simple harmonic oxcillator Vig) = (nvo/2)q? can be obtained as follows -Aly Zar Veterans = (2sim 2 (us) Now use the path integral method, as outlined in Problem 1.9, to recaver this renutlt in two ways: (a) by making use of the SHO result of Problem 1.8, then performing the inte- gration over boundary values of go = q(t,) = q(t.) as a simple Guussian integral, or (b) by using the approach of calculating the puth integral as indicated in Problem 1.6: Z cx det A~!/? with A being the appropriate operutor for the SHO ease. Solutlon to Problem 1.10 (w) Frown Problems 1.8, we have obtained the SHO transition amplitude (aretylquen) = | ——* | i Geil = | Tisinwt; — 0) P| 2sinwlty — 0) x(t + geno, —) Yara}. 116) “To get the partition function by following the method given in Problem 1.9, we set 44 = 0 and gy = q1 = qo: the exponent in the above equation becomes (Md 942 coment, = 1) ee = mMsINM ty /2) 9 Bein eats sin ety meosin(eoty2) 9 wy ey a2 = ~imeog3 wn SE au?) By integrating over q, und setting ty = if). we then obtain wa zn (—te I a. (ator, Aayexp| — (imatun SF ) gf 7 mee \"( . \*= (une 1 Faisinat,) \imientanor 2D =( ae = (oun), (1 118)4 Field quantization VAL (b) We start with the path integral representation of the partition function as given inegn (1.114): [l(a , 2 za fieaow|- f°] 3(#) + " ° @ oy = fudoieo[-t [ deg (—E +08) a] = NN" det Ay? (119) where A = —(d2/dr?) + 2 and det A = [], aq with a, being the cigenvalucs of ‘on the space of periedic functions f(r) = f(r + B). We expand this periodic function as f(r) = So, cn ("**7" then the eigenvalues are ay = w? +4772n2p-? for the eigenfunction e!*°®"', In order to evaluate such a determinant of a series, we first take the logarithm of the determinant Indet A = DY inte? + 4n7n?p, (1.120) To evaluate this series, we note thut Inder A fra axis =ofx ay ees, (1.12) In the integrand we have er woby x. d 3 2 2ptgndy-t wana A - do + 4nntp?y (1.122) $0 = x42) 08 + arnt BE (ey £ 2x = ath cah (ot). & coth SS Qn where we have used the identity ycoth my = 1 + 2y? Freya (1.123) Inthis way 7 Inder Aw [fay eth EX = 20 (sin 8X) 12a) or det A = (sinh(Bx /2))?, Thus Zen vin 82)" (1.125) which is in agreement with the result obtained in (a) when we Fecal that the temperuture independent constant in front hay no physical signifleance,LIL Non-standard path-tntegral representation 28 Non-standard path-integral representation Consider the Lagrangian with a position-dependent ‘muss* 1 Leal? (1.126) ‘Thus the canonical momentum is p = f(g)g and the Hamiltonian H = (p2/2f(q)). Show that the path-integral representation of the transition ampli- tude has the form ‘ : ot qa = NV fidarern |i fa [rae pao seo]}. (1.127) Suggestion, One cun start with the expression of the transition amplitude . Witla.) =¥ fda f unter: arin - ner = tim nf ee dq - dagen da ( 5 ‘) - (o.2+2-)]] (1.128) and then explicitly perform the momentum integrations. Remark. This is the counter-example, first given by Lee and Yang (1962), show= ing tat path-integral representation ix not always of the form (lat) 2M fidetesn|i f a La.@ | (1.129) Solution to Problem 1.11 Fort given Lagrangian, we have ’ 2 go w fal fines f ar [>a- #]\ stim ft... Boag... +dqn-t wal Soo (a2) - af The momentuin integrals are of the Gaussiun type °° dx exp(-At? + 8) = (t/ AY"? exp(B2/4A): af mole Ce) tall va ~ [Ae] p[ 2692 -a*] ; aa26 Field quantization 1,12 Since we are eventually interested in expressing the (cq | integrand in the form of an exponential, we will now write the position-dependent part of the prefuetor in the exponential form [,f(q)|"? = exp [} In £(q)}. In this way the above integral becomes: 2n\'? LOD (a= g-1)? _ in fad h (#) ox [22 (2582) ~ RL Toy) c.32) ‘This meuns that the path-integrul representation cun be written (gst lq.) = Jim conian-*? f dau dgnon Sf su) 7 ile few) xeon 5 [ A ( si tnd > =H fidarono|i [Ba - Ls] | 1.133) This is the claimed result. To get to the last line we have used the expression for Dirac’s delta funetion as ion 5(0) = fin, x 134) Weyl ordering of operators Notes on operator ordering For the simple system of which the Hamiltonian in the form P Qs 135 Hep.q) on +V(q) (1.135) hax no terms thut depend on both p and q, the quuntization is straightforward: just replace the classical variubles (p, 4) by operators (fi, 4). Thus the quantum Hamiltonian operator is unique: (1.136) and we have the puth-integral representation of the matrix clement hg GATHY Ipntan-w, Alb, = Len, ( Matta) lane), (1.137) (qe AGP. @lgr) f or Mm z For the more general case of which the Hamiltonian function H, (p. 4) ean have Inixed terms of 7 and g, for example p74?, the quantized theory is not unique, Each of the ch iad G2 p? will have the saine classical1.12 Weyl ordering of operators a limit, If these choices are non-equivalent, experimental measurements will, in principle, pick out the correct choice, Problem 12 will illustrate the point that for u Hamiltonian containing terms of the type q"p™, the path integral quantization corresponds to, in some sensc, the most symmetric ordering, culled Wey! onder ing. Wis defined to be the average of all possible orderings of ps and gs. For instance, a po le pot toon ms pd ) Tg" a pg + + pa") x a” pa! pg". (1.138) 2 24" (Paw ™ TaD os An instructive discussion of Weyl ordering ean be found in the book by Lee (1990). Remark. To do Problem 1.12, you may find the following identities useful: pe pent) (1) Yrs = qn +1214 1D (1-2) tat »o[! P yr =| gua (1-3) and Yes = (') (4) nant =>; ) 5) n(n = 12 Yu v(} ) (1-6) where (}) = (1!/1Gr — 1)!) are the binomial coefici identities? ts. Can you prove these ‘Suggestion. One approach to the first three relutions will be (© use the equalities Cheol! =D" = Lil = (r+ 1)" form = 2,3, 4. We note that the left-hund sides are just different combinations of Dfig/* & = 1, 2.3, The last three identities (1-4, 1-5, and 1-6) simply fotlow from the successive differentiation (a fdx")(b + xy" at x = | form = 0,1, 2.2H Fleld quamization 1.12 Problems on Weyl ord (u) Show that the Weyl ordering of operators jp und g can be writ 1 | " (egw a" + log tt pa) nt 7 resrtiad ‘ (L139) with the matrix clement of (gistl(pg" wlan ) (gisilplqn. (1.40) (b) Show the Weyl ordering of the operator product with two powers of p: gw a . 2 = potpgt=i™ rare al 1 Ft ytgt, ay which leads to the matrix element of (aera wlan) = (SE) aule*a (2.142) Remark. According to eqns (1.140) and (1.142), the matrix elements of g in the ‘Wey! ordering are just of the form (91,1 + q:)/2 8 prescribed in the path-integral formal Solution to Problem 1.12 (a) Before working out the situation for general n, let us first consider the simplest non-trivial ease of n = 2: (oP iw mG? p + apq + Pa) = pg? + iq (1.143) where we have used the comntutation relations lq. p] = img"! to move gs to the right of ps. The right-hand side is shown to be just the claimed result (ax given in the problem) by further application of these commutation relations: 1 I BP + gpa + pa?) = 5 [pg? + 21g) + 2¢04? + Ia) + Pa?) = pq? tig. (1.144)1.12 Weyl onlering of operators » ‘The matrix element then has the simple structure: 2 | 2 2 Mastl(pg? wlan) = 33 aistla"p + 2apq + Pa'lai) : lates + areas + a? gietlPland 2 + = (4%) (qetlolai). (1.145) Now let us repeat the above steps for the general n situation, (7q")w = ake pa") abe tig") H+ nel = pq" +5 (1.146) where to reach the last line we have used the identity (1-1), This result is the same: as given in the problem, eqn (1.139), because, ee 2 De Ten Od ae De Ty rs FY ls (‘) eet =i oa" + ity ROM fel a" + an (147) where we have used (1-4) and (I-$), The combination of eqns (1.146) and (1.147) yields the cluimed result; ene (1.148) rom, (paw = ao Po 2 y Na =D! The general tnutrix element can now be written as (qinil(pa wha) = (aeatla! pa") y! (i Veta villa oe = (234) (ausilplan. (1.149)30 Field yuamtization 1.12 (b) Before working out the generul n situation, let us first consider the simplest non-trivial case of n = 2: gw mw hq? + apap + apa + pa?p + pap + pq?) (1.150) ‘The right-hand side can then be rearranged by using the commutators [a". P”] imi" + pg"). ais) so that (raw = 4[ [0%a? + 2icgp + pad] + (cng + Dap +1c04 + pal + [0%a? + 2ipg] + [0%a? + tog] + 24" = }6p?q? + 12ipg — 3) = pq? + 2ipg - (1.182) ‘This last expression can be shown (o be just (gp? + 2gp%q + p%42)/4 which is eqn (1.141) with nm 2: 4G? ? + 2agpq + pq?) = 3 [(p2q? + 2ipg + 2igp) +2(p%¢ + ip + pg] = | [4p7¢? + Bipg — 2] = pq? + 2ipg - }. (1.153) Thus 1 aw = ner +2gp2q + pq. (184) and 2 ot + CaalteraPrwta) = (2222) cq tad. (158) Now let us repeat the above steps for the general n sit 2 . cpg" ie a TareD Loy! papa atm = 2 . ft fal) yom = ayes PT + qq" pg ine & singin 2 weber (og! pg?" 4 lglg) =1.12 Wert ondering of op ators 3 2 a . m1) gnctom = ETT Le [aloe +t ema Nt ea tna Fil [pglt"—! 4 id +m — Iyglt*-2] gt * 2 -GelmtD x D> [p?g" #421 + mpg"! = 1d +m = Dg"). (1.156) immo We will discuss separately the three terms on the right-hand side of this equation. In the first term we have the sum ao rye Dots he Sine n- Sr trad fat mad fa = (nt l= jntnt l) m jnt Dn $-2) (1.157) where the identity (I-1) has been used. We then evaluate the sum in the second ter of eqn (1.15 Leem =a Fis dm aS a0 im ead 1 223 fn-myin—m+ b+ Semin » ae = Syne + =m] sain + 0? = Pong 1) =o = prt 42), (1.158) where (1-1) has been used. To evaluate the sum in the third term of eqn (1.156), ‘We will need to use all three identities (1-1, 1-2, and 1-3); Yidem=n Kn=o - LiFims Lee SS = [He da -14 4+ -da-1+b] te 1g 3 m5 1-3? +3004 OP - (+2) 2B32 Field quantization 1.13 LP mt +1? n(n + DQn +1) n(n + 1) a5 TE tae 6 OPO = nin + Din+ 200= (1.159) Substituting the results from eqns (1.157) to (1.159) into eqn (1.156), we have (egw = pg" +inpg"! = yn = Ng", (1.160) ‘We next show thut the right-hund side is equal to (1/2") Dihg (")q! pig": = y (j)aerers 10 — x y (1) [etal + ing!" + pg] a" i = x > (?) [eta" + itt!" pg" + pg") i = ¥ > (') (v?o" + 2ipe"! 1d = Dg"). L.161 = ‘Using the identities (1-4, 1-5, und 1-6), we have i 1 z > (jeter = pig" +inpg"!— pata — Ng" (1.162) ra ‘Comparing cans (1.160) and (1.162) we see the relation of eqn (1.141) is satisfied. Generating functional for a scalar field ‘The generuting functional for the free scalar field (x) is of the form wasi= fuasierp|i fa'ecs] (1.163) where the Lagrangian density with an external ¢-numiber source J (x) is given by £1 = F09? — jd + Je. (164) (a) Show that such an Cy leads to the equation of motion (0 + 22) = J with a classical solution that cun be obtained by the usual Green's function method: OX) = fevers - yy). (1.165) ‘The Green's function (x — y) is the Feynman propagator for the scalar field in position space: dik aon "J amiBateie (1.166) Arte-y)113 Generating functional for a sealar field 3 (hy Show that, by a change of variable (x) = @-(x) + n(x) in the Lagrangian density. the generating functional can be expressed in the form wisi wero § [atettysenare- sun (2.167) where WV is a constant, independent of J Remark, One way to understand the ie prescription in the expression (1.166) for the propagator A p(x — y) is to note that the path integral expression in (1.163) is not well defined becauive of the oscillaory behaviour of exp {i f d'x £,} for a real £,. In principle, we have to go over to the Buclideun space-time = ir in order to convert this oscillatory behaviour into u damping one, We then return to the ‘Minkowski space by the method of analytic continuation. However, amuch simpler approach that will accomplish the same task is to add u term expli f d*x(/8)6"] with & > 0 in the generating functionul. This will provide a strong damping to the Gaussian integral. The generating functional will then be well defined, The Green's function for the corresponding equation of motion (O + 2? — ie)p =m J is of the form (Dx + 1? = i8)A p(x — y) = 34x — (1.168) with the solution as given in eqn (1.166). Solution to Problem 1.13 (a) ‘The minimization condition of the action (inodified according to the ir pre- scription as discussed in the Remetrk): Solem f ites = f d's(foo -fu2 seg +49] (1.10%) is simply the Buler-Lagrangian equation ay a (4)- (ui +J-OP =O. (1.170) ‘Thus the equution of motion is (Oy +n? = iP (a) = J. (a7) This ‘equation can be solved by the usual Green's function method ecr= fare-niord'y (1.172) With the Green's function (the propagator) being defined by the equation (Ox + 1? = fe) Ap(x — y) 84x - y). (1.173) Equution (1,173) can be solved by Fourier transform with the solution being given by eqn (1.166).ce Field quantization 1.13 (b) The change of variable @(x) = @-(x) + n(x) in the Lagrangian density of eqn (1.169) Silel= fatscs= f d's[joc-0- 1200+ 40) (1.174) Jends to an expression ofthe aetion in the form of Silden] = [es [-$9-0 + 079. + Jd] —nl(O + w)¢.- J1- fn +n} (1.175) Because ¢, satisfies the equation of motion, the second term on the right-hand side vanishes and the first term can be simplified, fet [-$o(0 +4209. + Jo] = fas [-doe4 + Jo.) =} f tsse08 w =-} fos JOD = Ody. (1.176) The generating functional being wal = ftorerotissien = fidniexpissie.+nl. (7 we can then factor out the part of action which is independent of the n(x) field: Wold] = exp {73 fos JOE -rsrd'y} x funnies |- fats nyo +c} = wero 5 fetrsenarte = nscne'y| (1.178) w= fuser -5 fas nena + neo} aay in independent of J (x). This is the desired result.114 Poles in Green's function 38 Poles in Green's function Consider the following generulization of the LZ reduction format: let T(q..-.) be a Fourier transform of the vacuum expectation value of a time-ordered product Tq. = fas 4@° OT (A(X) BCy) ++ )10). (1.180) where the operators A(x) H(y)+ + can be either elementary field operators like (x) or composite operators like ¢2(x). Suppose the operator A(x) has a non-zero matrix clement between the vacuum and some one-purticle state carrying quantum number a. (OIACO)Ip.a) #0, with (p? +m3)'? = E,. (181) Show thut in the upper half of the energy plane qe > 0 the function 7(q. ...) hus a pole structure of jim 7Q j= = AMP. a){p. a 17 (BO) -- 10) for gs > 0. (1.182) goon go mt+is for ga > 0. (1.182) Solution to Problent 1.14 We only need to consider in detail the simplest non-trivial case of two operators. Generalization to cases involving more than two operators will be straightforward. 'qQ= fas 1 (O|7(A(x)B(0))10) - [over [Oe {O]AC) B(O)10) + 8(—2){0] BOA (2) |} = fats : “fous Loiacono) + A(x) Loworararvcoroy) (1.183) 7 For the one-particle state |p. a) in {In)} r- late (184) We can use the translational invariance to write (O|A()Ip.a) = (OA) Ip.a) oP, and fos @P6(x) me (20'S a dt expli(go— Ea) th 2ny'5%4p — q)—_ = np — We Eat i@tes mi is = 28 - wz (1.185)%6 Field quantization 1.14 where we have used p = q 80 thut ao 2 mm? g-Fa _ WP g?=m pope oe Se . 1.186 2 Wo+ Es) Go+E.) — Go+ Ea) _ Inserting this into eqn (1.183), we can evaluate the function 7(q) in the limit of gq? > mi with go > 0 a’p 35% WE, te= | aero Toma x (O/AC)Ip. @)(p. a] B(0)|0) + = AMG. ana. a1B O10) g—mit+ie (1.187) where the ellipsis stands for the remaining terms. which are clearly free of poles and hence can be dropped in this limit, This is the desired result. We also note that for the ease of (q. a|A(0)|0) #0 there will be a pole term in the qo < 0 region of the forin forgo <0. (1.188) lin, 7 =Renormalization Counterterms in Ad* theory and in QED (a) Use the power-counting argument to construct counterterins and draw all the ‘one-loop divergent IPI graphs for the real scalar field theory with an interuction of (2.1) tb) Use the power-counting argument to construct counterterms for the QED. 1 ugrangian CaViytig— mv —ebytWAy- TEwF™ (22) where Fyy a By Ay = AvAye Remark. One of the key features of the QFD theory is that it is invariant under the gauge transformation (see CL-Section 8.1 for detuils) VOR) > WG) mH YX) bu + Pu =e Goo 1 Ayla) AY (4) = Ayla) + + atx). (23) ‘The desired counterterms must also be gauge invariant. Solutlon to Problem 2.1 (a) The superficial degree of divergence 2) ix related to the number of extemal boson lines A und the number of $3 vertives ny by CL-eqn (2.133): D=4. Bom. (aay () B= 2 (the self-energy diagram): see Fig. 2.1(a) for the one-loop divergent IPI graph, Thus D = 2—n}. Since the number of external lines iseven,.n, mustalso be even: my = 0 and 2, leading to quadratically divergent @? and logarithmically divergent 3.¢ 2 counterterms. (I) B = 3 (the g*-vertex di 2.1(b) for the one-loop divergent IPI graph. Thus D = 1 ~ ty = 0, ax my must be odd (hence m = 1), leading to a logarithmically divergent $* counterterm. (UI)_B = 4 (the o*-vertex diagram); see Fig. 2.2 for the one-loop divergent 1PI graph, Thus D = 0 leading to a logarithmically divergent ¢* counterterm,Renormatization 2.1 OO «x F OPy Remark. We muy also consider the “tadpole diugrams’ of Fig. 2.3 with B = 1 and P = 3— ny with odd , = | and 3, The quadratically divergent ones are shown in Fig. 2.3(b), while the logarithmic divergent one is shown in Fig. 2.3(a). (b) Here both the externul boson and fermion line numbers enter into CL-eqn (2.133) Dw4-B (25) Let ux enumerate all possible terms starting with the lowest possible external fermion (electron) and boson (photon) lin () F = 0, B = 2 (the vacuum polarization diagram, Fig. 2.4(a)): thus the degree of divergence D = 2 (i.e. quadratically divergent), In order to have a finite term weneed toexpand this contribution m,.(R) beyond the second order in photon momentum k: Fyv(k) = May O) + Bur (0) + kyukvAr(0) + HavlAd- (2.6) ‘Thus the required counterterms are (A)? and (8 A)?, But there is no gauge-invariunt counterterm of the non-derivative form (A). However, there 4s one gauge-invariant term of (8A)?: (8,4, — 8¢A,)(8"A” — aA"), which is the sume form ax the photon kinetic energy term F,, F#”.2.2 Divergences in non-linear chirat theary 9 w ) © @ Ne, 24. Uh) F m0, B = 4 (the photon-photon scattering tiagram, we have D = 0 (i.e. logarithmically divergent). Tysnothr) = Pyne 0) + Panap(kid 22 where Fyap(ki) is convergent. The required countertermn is of the form (A)*. However, it is not possible to construct such a term which is gauge invariant, Thus we would expect Myuap(ki) itself to be convergent. (il F = 2, B = 0 (the electron self-energy diagram, Fig. 2.4(c): the degree of divergence is one; hence it is linearly divergent. L(p) = £0) + PLO) + E(p) (2.8) where we expect £(0) to be linearly (or logarithmically) and £'(0) to be logarith- mieally divergent, £(p) being convergent, The required counterterms are E(O)W and 'O)yy"8,¥, respectively. (lv) F = 2, B mI (the electron-photon vertex diagrum, Fig. 2.4(d): it is logarithmically divergent, becuuse 2 = 0. Pyulp.g) = Ty) + Fyu(p.g) (29) with counterterm of the form Py" ¥. . 2.4(b)): here Divergences in non-linear chiral theory The non-linear SU(2) x SU(2) chiral Lagrangian is of the form 2 Leo £ Tr(a" ut a,U) (2.10) U wen (= ie and + = (x1, t2- 3) are the Pauli matrices.40 Renormalization 2.2 Also f = 94 MeV is the pion decay constunt, (For more discussion of non-linear theories, see Problems 5.7, 5.8, and 5.9.) Use power counting to enumerate all the superficially divergent Green's functions at the one-loop level und construct the appropriate counterterms, Solution to Problem 2.2 ‘The superticial degree of divergence is given by D=4-B+ YY nib +d, -4). Qu For the chiral Lagrangian, we have the number of derivatives inthe th vertex being d, = 2, for all vertices, Substituting the topological relation 2UB) + B= Yomi (2.12) T imo =a) — Dont (2.13) we get 1 B 5 mb -D-Z+l. (2.14) 2 x 2 ‘The superticial degree of divergence is then Dm 4— B+ Y milo —2) e242. (2.15) ‘This gives the result that at the one-loop level (L = 1) the degree of divergence D = 4, independent of the external lines. It implies that the number of derivatives in the countertermns should be four or less, The term with two derivatives is just the erm in the original Lagrangian, The four derivative terms should have the form (09) A" P,O)G"O)O". even, (2.16) Similarly for the other counterterms, Taking into account the SU(2) x SUQ) symmetry, the counterterms are of the form, [rr(a"ut a, uP Tr(@" Ut ,U BU" AU) Tr(@*UUtA,UU! a,t/Ut aU") Fr@*Ut aU) Er @U* aU) Tr(B"UUt UU) Fr(8, UU UU"). (2.17)2.3. Divergences in lowerdimension tl field theories 4 Divergences in lower-dimensional field theories Consider theories with scalar field ¢, fermion field y, and massless vector gauge ficld A,, in a d-dimensional space-time (d — 1 space coordinates and | time coordinate). Express the superficiul degree of divergence of a diagram 1) in terms of the number of external fermion #, boson B lines, and the number of fermion fi, boson 6; lines und the number of derivatives dj in the ith type vertex. Namely, deduce the generalization of CL-eqn (2.133) to ad-dimensional field theory. Keep in mind that the propagator in d-dimensional field theory hus exactly the sume form ax those in our physical four-dimensional momentum space, For example, the propagator for the gauge field A, in the € = 1 Feynman gauge is aR Brie Use the formula deduced in (a) to write down all possible renormalizable and stiperrenormalizable interactions for dimensions: (i) d = 2 und (ii) d = 3. iy) = (2.18) Solutlon to Problem 2.3 1m the structure of the graph we have the relations B+20B)= Somb, PF +20F)= Sonifi. 19) as well as (U2) + UF) Son els (2.20) where L is the number of loops in the gruph. Fhe superticial degree of divergence can be culeulated by using the relation (2.20) D =dL - 208) - UF) + Yond; =d (um +0F)- Som : + 2B) = (UE) + Yo ned, (2.21) 7 7 Eliminating /B and IF by eqn (2.19), we get De (4) [>a - | + + Yond - ata ns (B) (a2 Renormalization 2.3 Or d-2 d-1 Ded (Pye (PF) redo (2.22) where &; is the index of divergence, a= (=) at (‘) fitd-d. (2.23) Clearly, these results check with CL-egn (2.133) for the cuse of dimension d = 4, () The d = 2 case: Equations (2.22) and (2.23) are reduced to F fi Der zt Yinde bateyn2 (2.24) © super-renormalizuble interaction. 8; < 0 e, nw3,4, .. 5(g") = -2, Ve" neh. OD Vrwa", — 8byWA") m1 (2.25) Note the interaction of the type Ay A” is super-renormalicable but not guuge invariance, © renormulizuble interaction, 8; = 0 en dml PyaWayde", Prey wFe, a =2 ga" g)9". (2.26) (I) The d = 3 case: Equations (2.22) and (2.23) are reduced to D B b ~ zh Don wade tth-3 (2.27) © super-renormaliz tble interaction, 8, < 0 dm 6 OO drat, d=l. ddA". (2.28) © renormalizuble interaction, 5; = 0 dad, dyer. Gal adatg (2,29)2.4. n-Dimensional ‘sphericat” coordinates 43 Dimensional ‘spherical’ coordinates Inan n-dimensional space, the Cartesiun coordinates can be parumetrized in terms: of the ‘spherical’ coordinates as 1 = ry 8101 SIN a2 2 ry Sin Gy 8in 2 +++ Sin 8 CON}, 3 = ra Sin By 1 sin Oy Xa = Fa COS By where Osh s2m, 056.6. and Show that the n-dimensional infinitesimat volume is given by Hsin Oy=1)""2 (sin Oy2)"9 ++ x (sin @)(d0, dO +++ d0q-1) dry (2.33) dx dxydxy...dXq assed in CL-p. 53. Solution to Problem 2.4 We will solve this problem by finding the relation between the volume factors in and n — | dimensions. Namely, we will proceed from the simplest n = 2 to higher-dimensional cases: (n = 2) —> (1 m3) —> (n = 4) —> (general n). (2.34) (a) n=2 Here the two Cartesian coordinates (x1,.%2) are related to the familiar polar coordinates (rz, 61) by xrsrsing, xm rcosh, 050 s2n, aapts} (2:35) The distance dsz between neighbouring points is given by (ds)? = (day)? + (x2)? = (rz)? + Fd, (2.36) ‘The ‘volume* clement is then the product of segments in orthogonal directions, i.e. the product of the coordinate differential with the appropriate coefficients ox indicated by the (quadratic) distance relation: dV, me dt dxy oe (dry\(rod&h) me rodeo dth. (2.37)44 Ren wmatig tion 2.4 (b) a m3 Consider a sphere in three dimensions. If we eut this sphere by a plane perpendic- Ular to the x-axis, we get a series of circles in the planes spanned by Cartesian coordinates (X41, 42) which are related to the polar coordinittes (r2.0,) by x =rsing., x2 = cos, 0<6 2m, rpextexd. (2.38) The infinitesimal distance on this plane can be expressed in these two coordinate systems as (dxy)? + (dig)? = (dra)? + de)? (2.39) We can alsocut the sphere by a plane containing the.xs-axis, resulting in a series of vertical circles’. On these two-dimensional subspaces, the Cartesian coordinates ure (r3, x3) and the corresponding polar coordinates are (73,62). We recognize that, 4 is the usual polar angle. nensind, my =ryco, 0 S&S. (2.40) Hedtxpoxttad tah 2.41) As in eqn (2.39), the infinitesimal ways: istance can be expressed in two equivaent (dra)? + (x3)? = (dry)? + Fae)? (2.42) Combining the two sets of coordinates in eqns (2.98) and (2.40), we get the usual spherical coordinate relations, x = rysin 6 sin 0, x2 = ry sin ® cos}, X1 = ry con. (2.43) We can turn the distance formula (sy)? me (dey)? + (dea)? + (dy)? into (dy)? = (dr)? + 73(d6,)? + (xy)? (2.44) by using eqn (2.39). This can be further reduced, by using eqns (2.40) and (2.42), t0 (day? = (dy? + (dey? + av? = (drs)? + 73 (do)? + dey? = (dry)? + 12 sin 63(d01)? + 7340)? (2.45) ‘The volume clement can be obtained from the product of these three terms Vs = (cdrs)(ra sin 82 dO1)(r5 dO2) = 2 xin 82(dry dr d62)- (2.46)24. mDimensional ‘spherical’ coordinates 4s nad We can also imagine cutting the sphere in four ditnensions by the manifuld (x4 = constant) to get a thtee-sphere with radius r3 = rasin @, where we can introduce three-dimensional spherical coordinates (r3, 01. 02) ax in eqn (2.43) &1 me ry sin 6 sin = (rg sin G3) sin & sin 6, Xz = ry sin 8; cosy = (ra sin &) sin @: cox 4, Xq = ryCOn Bz = (ry sin 6) COs 62. (2.47) with the distance formula (2.45) (dx)? + (dxz)? + (dx)? = (dry)? + 72 (d62)? + 73 sin? (401), (2.48) Now we introduce two.dimensional polar coordinates (rs. 63) in the (rs, x4) plane ry =rasin Oy, x4 me 74C08O3, (2.49) with the distunee formula (drs? + (axa)? = (dra)? + 12 (d05)2. (2.50) Inthis way the infinitesimal distance in this four-dimensional space can be rewritten by using eqs (2.48) and (2.50) (dsg)? = (Cd)? + (dg)? + (dx3)?| + (dxa)? = [Mdra)? + r}(de:)? + 73 sin? &(a01)"] + (xa)? sm (dry)? + (dxs)?] + 73 (d82)? + 73 sin? 6 (d0,)? = (dra)? + 73 (03)? + 73(d0g)? + 72 sin? (40)? (2.51) The infinitesimal voluine clement is then = (drs)(rg 40s ry dO) (ry sin Op dO) r} sin? 6) sin 6,(dr4 d6, dz ds) (2.52) where we have used eqn (2.49) to reach the last expression. (d) General x= in Oy -1 Sin Oy.2 ++ » Sin & sin By, 2 = ry SiNOg 1 Hin Oy + +8) COBBL, 3 ry SING Sin Oy +++ SiN By CORB, Xn tn 6O8 Inat (2.53) where raxtgxdte teh Os ys 2m, 056.0%4 Renarmatization 2.5 The infinitesimal distance is (dq)? = (dx1)? + (da)? +++ + (dy)? = (drq)? + r2(dOq-1)? + 12 sin? Oy-1(d6,-2)? + r2sin® Gy.1 sin? @,-2( 109-3)? +-++ + rP sin? Opt sin? O,2 ++-8in? (401)? (2.55) und the volume element ix AV, me dt dys (rn) (sin Oy 1)" (sin Ba)” ++ x sin O2(drq dO, dy +++ dOq-1)- (2.56) ome integrals in dimensional regularization Use the dimensional regularization to derive the following results for the Feyninun integrals with denominator power or in n dimensions: (n) =1)"? (a — nf2) 1 _—$<$<——=$s ———— an Gm Fa) Ma paiayeom (257) ak k b) y= ff —————— ” Man) = | oe Gia ap k+ MPa lee = = pulolann), (238) ak Kak, Ce © alam = | Oly bap ke MPa Te 1 M?— p? = <8 |. 19) Ila.n) [nee + Goanep} 2 Kyukok, @ Iyvol@tsn) = fa a aren) = | Oe Ge ap kt MEET = hie. [Perens + Buc Pe + Bue Pe + Rep Pa) we “@onh-D}2.8 Some int grais in dimensional regularization a7 Solution to Problem 2.5 (a) In the Feyninan integral d"k 1 totoom= f Gry aap b+ Mee 2.60) the denominator can be written as Dw 42p-k4+M +le= (pth? + (M— py) sie =k7 4 (M = py tie (2.61) where k’ = k-+ p. For the case p? > M?, we can perform the Wick rotation to get D=-P 4M - p tie =-(P +07) (2.62) where a? = p? — M? —ie and dk 1 1 Jola.n) =i (2.63) Gay Ge +a As usual [see CL-eqn (2.112)), the 2t-dimensional angular integration gives nia 2 fem "Far (2.64) m1) ann? po lak Gay TH) bo Fa (ety an |p a "Gry Pupil wre (N78 a? 1 Pinf2P@ = n/2) Ina.n) Gay Ped) (aye T@) (-ny'? P@—n]2) 1 "Gar Ta) OP paras A One of the most common convergent Feyninan integrals has o = 3, dk 1 h3n)= | ——>— SO m= | Gay pap ke ME ee (=n)? TG = (@/2) 1 “Gar Ta Wiaprerepae (66) which gives, forn = 4, dik 1 CC ——— net Sam OF Ip-k+ Mi + iy i 1 “Sura (2.67)48 Renormalization 2.8 dk ' — 4+ __——. (2. ry hem = | oo Ge Ip ae a Te (2.68) Ax before, we set k! = &-+ p. Then ny Ken hee | oe (2.69) Caryn (K+ aye” and the term lincar in kj, gives zero because of the symmetric integration, The result ix Ila, n) = —pylola.n). (2.70) (tetany f fate o (ny (K+ 2p k + M+ Ley _ f atk ky — Pulky — Pod “Gry +atye dk _kyky vk ot ne =< 2.7 Gry eae + | Gare eae 7) In the first term we can replace ky ky by (1/m)k? Quy to get [ dk ky Vf ak Qny ade on J Oxy +a af [Ok Rta n-a aS Qry (+ a2" = BE [ite 1m—attola.m] (2.72) Using the identity P(x + 1) m xI"(x), we get = bet elt a ete.m) 7) ela — 1. = ay and I Tyla.) = [oom + fe lM? ere Ian). (2.74) For the cuse a m4, we huve i, ak he Iuvt4ny me f SK inks wet) Sar (4 2p k++ I = [rom + $8 uvlM? rhe | fot4.n). (2.75)2.6 Vacuum polarization and subtraction schemes 49 which gives fora = 4, ‘ Tev4.4) = [Pure + del? — p7I] one? ta (2.76) htm = [fk kykikp wo nel | One lp k+ Mi + ee fk ki Pudlhy = Pky “J Qa (2 + ate Dropping terms with odd powers of k’, we get d’k 1 Lup) = Tay a ap kee + kyu poke + Pukokp) — PuPvPph = —PuPv Pylon) = (P slaw + Pulyp + Prtap) we = hte, [eae + J (RurPp + RupPv + ror orl Vacuum polarization and subtraction schemes Use the dimensional regularization to compute the one-loop vacuutn polarization in QED, Solution to Problem 2.6 ‘The usual vacuum polarization in QED ix given by ing) = gg — Pe ing?) mies)? f 8 re| ye ‘ieo) faery con In the dimensional regularization, we can replace d‘p .f ap ane J Gay a where 8 = 4—d and is some arbitrary mass scale that one can introduce in the dimensional regularization scheme. The integrand can be simplified as -rrf[ye—t__ep_i___ wr» Jonah” j=] 1 1 = [ari] [i msi ] Try + myPp — 6g + (2.80)50 Renormalization 2.6 ‘The numerator is of the form NF = Try By Am THy*y*). 2.81) ‘Dirac algebra in d ditnensions gives 2g 1, (2.82) where /; is the identity matrix in d-dimensional Dirac algebra space, with the truce 7" Trl, = f(d). (2.83) Here f(d) ean be any function so long as it has the property (4) = 4. It is straightforward to sce that Tr BEA) = Sida + bye +d) — (ab -d) + (a-dyb-e)}. (2.84) Using these we get for the numerutor Ne = f(d) (Ip*(p — 4 + Pp — 9)" — gp (p— gl + mg). (2.85) ‘The denominator is calculated in the usual way, (2.86) where A= (1—ay(p?—m?) + al (p — 9g)? — m7} = p? —2ap-q—m? +aq? = (p-aqy a? (2.87) with a =m? — al-aq? (2.88) The vacuum polari ation is then of the form fp yee f ———____ _ (2.89) Ory Hy page atl To simplify the integration we shift the variable, p > p-bag. Then the numerator (2.85) becomes P(g) = ied af da N* > f(dy{\(p +ag)"(p = (1 — ag)? + (p +ay)'(p = (1 — ag)" = RF lm? — (np aq) -(p— (1 arg) = f(d){2p* p? — 2a(1 - adg*g" tat lm? — (p? — act - @d9")1} (2.90)2.6 Vacuum polartzauion andl subtraction schemes SI where we have dropped terms linear in p which will vanish under the symmetric integration in p. Then we have dt 1 ni (y) miedul af af oom eae a er x [208 9? — 2a — engtat + etn? ~ (p? - at - yh). 91) “The term proportional to g* can be written as Hp? = m2 + (1 = erg? | + 201 = ag? = —(p? — a7) + 201 — oq? so we get 4 dp | 2ptp® a1 —a)y2y? Hg) = Oe | tee a ae nh(q) sien Sid) f da f Gay |G@oa? ~ Faat? joe oe | ee (aay paa d'p 2p%p* gt sien! reo da aa [aoa Sr dail = ar(g?a®? ~ 4°4*) er o| The relevant formulue for the dimensional integration are eqns (2.57) and (2.59). (ea? Fr {2} — (2. Fa) ea _ a= at) =f” (2.94) Using these we get .LlUD UD rt—*“‘#ECEC Or ae ~ 1-47) Ont FQ) Eanes ap et geni my¥PT=d/2) 1 Gn P= Qayt TW) (296) Using the idemity (297)2 Renormalization 2.6 {3 [aaa an l= oF Now the vacuum ae is of the gauge-invariant form, aloo) para — ate) f der in < we can show that (2.98) wR (g) m edn! fd) aun im? all — ang Expanding in the power of «. we = 2 —y +In(4z) + Of) (2.99) where y = 0.5772... is the Euler constant, wha"? — 1 = Sina? = Ing?) (2.100) we get apy pt ed tq) = LO Pda ay {2 = y + dy — nf Br Jy 3 Write LA) = fAVF- 41°) + A Hae tees) (2.101) where a = —$,/"(4). Then we have rg) = ZI ! [2 —y+in(an) +2] 2 a) ang? -{ deat ~ayn[ AE), 0 H Different subtraction schemes From this we see that different choice of f (d) corresponds to different constants in the finite part of x(q), which is arbitrary anyway. For convenience, we can choose a= 0, or f(d) = 4 forall d. Minimul subtraction scheme (MS). Here we subtract out the pole in & to get gil muss(q) = 505) g(-¥ + Ina) n -f daa(| —a)In ewe lo ‘This corresponds to choosing the renormalization constant, 2 (23s = 1+ 3 (2.102)2.7 Renormalization of 49? theory in n dimensions 33 Modified minimal subtraction scheme (MS), The renormalization constant is chosen so that the term (=y + In 477) is also removed from the finite part, aot 2 2 = _ am? — af) — og’ ust) = 53 [ daa(l ayn nee noe : From this rexult we can study the low- and high-energy behaviour of x(q). For lg? «m, (2.103) : 2 2 In [= oe a? ]=m% aa oS (2.104) ff eat ~anin[ 20 O | ho Kw 1 2 a -[ deatt [05 ~ at -o 5] lo # mv Feuer A as =} =-H(S)+ (2.105) Then we have: dm, i (@ manta) = Sgn + (S) tee, (2.106) In the other limit Ig[? > m?, ‘m2 —a(1 — aq? ~a(1 —a)q? w[=aee x(t aac +] and 1 ang? feo -orn[=— or) lo Hw =?) ft nl =(=) f daatt—ar+ f daa( —a)Inla(l — a)! we) So o (2,107) #)-3 ‘Thus the limit is 2) se santa) = ~ 7b n (2 \+ Re (2.108) ion of Ad? theory in 2 dimensions ‘Consider the A¢* theory where the Lagrangian is of the form enormali £ =H ae ore me - Be. (2.109)4 Renormati tion 2.7 (a) Show that Xp hax dimension (6 — )/2 in n-dimensional space-time, To have 2o with fixed dimension for arbitrary n, we can define 5 ama | wh ee tt Fo= mun", Rodan’, with @= (2.110) where 1 is an arbitrary mass scale, (b) Show that the one-loop divergent graphs for the case n = 4 are those given in Fig. 2.5. (©) Carry out the renormalization program for this theory by using the MS scheme, Solution to Problem 2.7 (a) Since the action S = f d"'x L is dimensionless, £ has the dimension 1, From the mass term or the kinetic energy term we sce that @ has dimension (n — 2)/2, which gives the dimension of Ay as (6 —)/2. (b) From eqn (2.23) we know that the index of divergence for the $? interaction is o('F)-(9) am which, as expected, is just the negative of the dimension of Xy. The superficial degree of divergence ix then ben ("Z)a40 2.112) where v ix the number of @* interactions in the graph. The number of loops L is given by eqn (2.14) L=fw-By+l. (2.113) Thus in one-loop we huve v = B. and D=n-28, (2.114) and for four-dimensional theories | = 4, only B = 2 self-energy and # = I tadpole graphs are divergent. —C)— ha, 2.8, Solf-energy and tadpole dlageamn in the 46? theory.2.7 Renormalization of 46° theory inn dimensions 5S (c-l) The self-energy graph ony i i 1200 OE | Sea ND Itis straightforward (o evaluate this integral to get ee da 2 Gay Jy im?—al—ayp?—iel (2.116) Uning Te=t-y +o, df alteina+ Oe) (2.117) we can expand £(p) around € = 0, to get 1 ft , dry? ro=22,) -v+n (SE) ft ac [= -a(l-a)p?- te] | ly me We now rewrite the Lagrangian as Joa gj? — Mtg? — gs — Lgntg? C507 7? a? zane (2.118) with dm? == on? — mi. This amounts to the replacement m? —> mj and add the term 45m¢?as a new vertex. The new self-energy £*(p) is then EAX(p.mg) = E(peme) + bm? (2.119) Now if we chooxe 3 0 jt = Bolg te (2.120) where cy, is finite for ¢ > 0, but is otherwise arbitrary (different renormalization schemes correspond to different choices of cm), then the pole at ¢ = 0 cancels ‘out and e 4 Epa) = Sp [o- - , dain [z lo Is finite. ‘We now study this choice of ¢,, in various renormalization schemes,56 Renormalization 2.7 ) Momentum subtraction Suppose we choose Lp. ma)lpte-a = 0 (2.121) for some M?, (This corresponds to normalizing the propagator such that itis the ‘same as the free propagator at p? = —M?,) Then the constant cm is given by a on Sg asi eye (Se) +f dan [Bat eO Seyi aie oom ‘| (2.122) me) * Jy mie, and the self-energy ix \ 2 2a R aol ma —a(i—a)p?—ie Pamd=5 Saal datn| EES aM? 16 |" dy) Minin Cm = 04a out 24 amit 5 Oly rpm) = Bo y+in(2") tO Oct hee me 2 oy ' ‘nt, — a(t — a) p? — ie eye” f ean mice serat (li) Modified mininum substractlon (MS scheme) It tums out that the combination (2.123) subtraction (MS schetne) Thix corresponds to the choice nouns that we subtract only the pole ut & = O und the self-energy is (fs) = y + In (2.124) always appears in the dimensional regularization, Thus it is convenient to choose cm =y~In4x (2.125) (2.126) (c-Il) The tadpole dlagram ‘We have (ant Cat ft See ie hing! n ar Rak 2a Dem". 20 This will contribute to the vacuum expectation value (016010) = 5 (2.128) Es ive an infinite constant vacuum expectation value for the field. To clim- inate this infinity, we cun udd another counterterm to the Lagrangian of the fort Lad = =. (2.129) This counterterm will huve the effect of cancelling all the mdpole terns, without ierfering with any other consideration,28 Renormattzation of composite operators 7 Remark. The parameter of renormalized mass, m g, can be related to meuxurable ties as follows. ‘Ihe physical mass of the ¢-particle, mp, is defined to be N Of the single-particle pole in the two-point funtion, This gives my as «function mp = mn glnRs A oC) (2.130) We can solve for mg in terms of other parameters, 1g MARL IMps Ry Ha Cm)» (2.131) tis important to note that the parameter mg hux implicit dependence on the arbitrary mass scule je. enormalization of composite operators Consider a theory with a fermion and a complex scalar field, The Lagrangian is a = (dO) — 178? - 5° THY", — NW + BOO + hee. (2.132) Show that the composite operators Osby" — OF = 1G" yb - 649") (2.133) mix under the renormalization, jon to Problem 2.8 Vertices for the composite operators are displayed in Fig 2.6, where solid lines are fermions and dashed lines are bosons. Add two terms to the Lagrangiun for these operators Le LAI XO) + i day VOR (2). (2.134) {he one-loop divergent diagrams for these composite operators are shown in ig. 2.7, DX a CEES “ ON ete bye 2. Z ~ - . Tub SH v. ce N38 Renormalization 2.8 oF af \ Pt co) of of x A LN it as os, af NA Ae “A © () Fo. 27. Figure 2.7(a) is a logarithmically divergent Green's function for the operator Of with two fermion external lines. We can separate out the divergent purt as PE Can pa) = y"E (0,0) + pisP ++ (2.135) where the first term is logarithmically divergent and all other terms ure finite. Thus ‘we need a counterterm of the form al OF YY SylX) = IPO Oy OF. (2.136) Figure 2.7( b) gives « logarithmically divergent Green's function for the operator O* with two scalar external lines. Again, separate out the divergent part as Tisha ka) me (ky + YT 9 (0, 0) + ++ (2.137) It shows the necessity of a counterterm of the form iT 90,04 OF. (2.138) Figure 2.7(c) shows the necessity of a counterterm of the form iT 2c(0, 0) Joy Of (2.139) Figure 2.7(d) shows the necessity of a counterterm of the form —iT2y(0. 0) Ja, OF. (2.140) ‘Thus the effective Lagrangian which contains the composite operators and their one-loop counterterms is of the form Le = iSpy Ot = 0,0. 0) + iy OFC = P2400) iS, OFT (0.0) — iJ, Of Tie, 0) = iy OU Zu + day OF Za t idiy OF Z12 + bau OF Za, nit om of =1ta (2 28) (ch) =15!'2,0% 141)29 Cutkosky rules 9 where Zn =l-P 0.0), Za -Py(0,0), Zn =t + P24(0.0), Zu = —Tre(0.0). (2.142) ‘The renormalization constants are now in the matrix form, explicitly display the nixing of these operutors, | (20 2a z= (3 2) (2.143) which ix neither symmetric nor real. Nevertheless we cun diugonalize this by biunitary transformation (see CL-Section 11.3 for the details of biunitary transformation), Z=UzVt (2.144) where: 6 2) (2.145) is diagonal, V and U are unitary matrices. Then we can write Lo wid! 20% iS! 2,0 +i Ff 2,03 (2.146) where 6" evi jteut Of aVjof. Tt wuyst. This means that neither Of nor OF are multiplicutively renormalizable. But the combinations: Of VOL + ViOF. OF = VO! + VQ OP (2.147) are multi ively renormalized. Cutkosky rules In the Ap* theory, the one-loop diagrum in Fig. 2.5 gives the contribution i wp atk 1 ! ot Oars | aipse—teniea ay eM Show that in the complex s plune, the imaginary part for s > 42? is of the form It fats) = ZIPS +18) — 1s — de)] Moe dk =I aye 2778p + hy? = WBE? = 4), (2.149)60 Renormaliz tion 2.9 Solution to Problem 2.9 ‘rom the structure of the two propagators, the poles ure located at the following two locations in the complex ky plane. From the second propagator, we have ko = £ (2 + 42)!? zie = £(Ey - i) (2.150) where Ey = (k? + 22)'/2, and from the first propagator, we have ky = pot (pth? + 2)!? Fie = pot (Ep ie). (2.151) jegrand is of the form 1 1 1 te * Go Ey +16) Oot Ex — 16) at po + Eps — 18) 1 “Got Po E pax tie)” ‘We can close the contour in the upper half plane and get the contribution from the residues at ko = —Ey + ir and ky = —po — Eas +i8. () Residue at ko = —Ey + i€ : : - hs tnt Ex — 16) (—Ex + Epes + Po — 16) (Fa — Epes + Po Fie) (2.152) The lust two terms in the denominator can be put into the form (Ex po? = EX = [tk +29"? — pop w=? = p?-2mEr, —2p-k tie, For convenience we can take p" = (pp. 0), then po = and 1 21). 2.153) = SEL eo ' Since Ey > s2.we see that for § > 4p? the factor (S — 2VSE, + ie)~! has singularity along the path of integration (from Ey = j« to 00) und will contribute to the discontinuity for $= 4,0, (I) Residue at ky = —po — Eps + it hh = -2nt ie)" (2.154) (2B pst In the denominator, we have (po + Epsa)? — Ete = ph (p ky? +o? + pak pas — 2 =p? = ph +2pok’pis (2.185)29° Cutkasky rues 6 and : ' 1 nets (s0:) ape +e" 156) Wis easy to see that this denominator never vanishes fors > 4,22 und will not give discontinuity in the physical region. Thus if we close the integration contour in the upper half kp plane, only f; will contribute to the discontinuity in the physical region. For the calculation of the discontinuity, we write 1 I —Cn( |) 2.15 A can (5) (2.157) P(p — 2E, + ie) where p = J. Using the formula I x-axis xna@ Fir8(x —a) (2.158) ‘we can obtain the discontinuity across the cut for the cuse s & 412, 1 2 disc h = h(p +is) — h(p — is) = (-2ni7'8(p - 260 (ze) © (2159) To get a more systemutic rule for calculating the discontinuity, we write /y as Perey Se Bree Pel 160) (reo tis 2.160) he fe (—27i)8(e = 4) ‘We sce that this corresponds to replacing the propagator (k? — y? + ie)! by (271) (K? — 22). Similarly, disc) = fate (-27i8? = 2)(-2ni)5(p +H? 7) (2.161) or a dura. = | 28 ; Gaye AR BE? — se Da NTC + AIF = 7] (2.162) which ix the requested result. Discussion As indicated by this calculation, we cun obiuin the discontinuity of 1°(s) by putting each particle in the loop on the mass shell with the replacement of Fea pt ate (- 2015? — 1) 9(4). (2.163)62 Renormalization 2.9 In fact, this discor ‘we now illustrate: dks dhe ; (Gry Qn ity can be written in terms of physical matrix elements. us 1 84(k @oein Basin’ ag (288k + ka — P) Ty t ie) 18 +e) res Then uaing the replacement given above we get the discontinuity as di ney a Bl f tht oe, et f San (taste = 220Ch) aka 2_ 2 2n)4 xf Spt 2M 18 — IQ x 54k + ka — p)O(k20). (2.164) The integration over ko is of the form {? (285 (K? — 12? )8 ko) = (2.165) where Ey = (k? + 47)", and vise Ps) = 5 dk iatomtst aa f Tarte CP ante + ha - pr. ‘The factor (~/A) is just the scattering amplitude in first order of A, 7, == -iA. Thus to order A?, the discontinuity of the scattering amplitude in the variable 5 == p? ‘can be written as the integral over the phase space of |71[?. This ix the essence Cf the unitarity of the S-matrix, SS" = S*S = 1 which implies for the T-matrix (Sm 1+iT) T-Tt=7T (2,166) or LMT n 167 ‘The prescription of replacing (k? — 2? + ts)~' by (—277/)6(k? — 12) is a simple example of the Cutkasky rule which gives a general method for computing the discontinuity for an arbitrary Feynman diagram and is summarized below: ‘© Cut through the diagram in all possible ways such that the cut propagator can be Put simultaneously on the mass shell for the kinematic region of interest (e.g. only for s > 4.2 can both propagators be put on the mass shell). © Foreach cut, replace the propagator (k?— 2-4/8)" ! by (—27i)8(k?— 127 )0(ko) und perform the loop integration, ¢ Sum the contributions of all possible cuts.Renormalization group Homogeneous renormalization-group equation Consider the Ag‘ theory in d-dimensional space-time, where the Lagrangian for = 2— (d/2)is given by ' : Cm 5 oy — Tg? — yt 2 ' Fr dns = 5 On) - Be ~ Fi oh + (counterterms) Gl where Ag. da, and tng are renormalized quantities, und yz is the urbitrary mass scale one needs to introduce in the dimensional regularization, Use the fact that the unrenormatized -point Green's functions P (p;, Ao. mo) depend on the bare parameters (mo, Ag) and are independent of the urbitrary muss scale, f4, present in uny scheme of dimensional regularization, a HP (Po Rov ma) = 0, with may Ao held fixed 32) to derive the renormalization group (RG) equation for this theory. Solution to Problem 3.1 Recall thut the relation, CL-eqn (3.50), between unrenormatized and renormalized Green's functions is given by TF (4, do. tad = Z5"7 PO (pi Aes e i) (3.3) Thus the statement of 424/420 = 0 means that ee [2a (or Anvine. so] =0. oa Note that both A and mp depend implicitly on zz, Thus we have nn @ a dhe @ Oma @ [jeu ote aa ae et | xPE" (Pr Aremas a) = 0. Defining the quantities, 1a aa am = sua InZ,, sua aut. YOu = 7Hg, Ze BO sua. ya (hema ag Q5)64 Renormatigation group 3.2 we can write the renormalization group equation ns a a 8) pum . [uz +B (Ar) met” Og) + MR m (An) z;] Py (Pde R = 0. (3.6) Remark 1. Implicit in such calculation is the fact that the bare quuntities 4y and img are held fixed, Remark 2. ‘This equation is w homogeneous equation which ix more convenient to work with than the original Callan-Symanvik equation. Renormalization constants In Ag theory, the renormalized and unrenormalized coupling constants in the dimensional regularization scheme ure related by AR) = WO Z"yeddo (3.7) where Z is the coupling constant renormalization constant of the form (1 =Z0e, (3.8) where Z, and Zy are defined in CL-egns (2.23), (2.36), and (2.40), (a) Show that the f-lunction can be written as dz Ag) = -ehg— TAR 39) ie (b) In the one-loop approximation, we have = 3a Zist- lente’ Zy=1+0(¥). (3.10) Show that a2 e N-S : 1 BO) Tort +O (2") GAD Solution to Problem 3.2 (a) By explicit differentiation of eqn (3.7), we have ar Aan =H5E a s “On (u*Z(u)ro) . d me eu 2 pyro — nM "a dz = whe Bae (3.12)3.2 Renormalization constants 6s (b) Substituting the one-loop result (3.10) into eqn (3.8), we get 3dr T6r%e G13) g2 3 the 3.14 "Gn ~ Vonte de Teehon a wring higher-order terms, we obtuit 3 Text P (An) (3.15) way the relation (3.9) becomes we a, 3.16) BAR) = FAR - Fan Are neh EO. (3.16) Solving for f(g). we get Baw ( Bae Bs) — = : 3. pan sha ( t+ isn pty = ede (1- GE In Taking the limit ¢ —> 0, we obtwin the stated result: pane BE 400%). (3.18) Remark. More general atulysix of dependence of 6(A) can be cartied cut as follows, We first write equ (3.7) ox doa Au'Z, (3.19) From jeddo/dpe = 0, we get 5. do. 3 ds HM eO7)+u—(Z)| 20 or eZ) =p— (AZ). (3.20) dp dp In the MS scheme . 7 can be written as power series in (I/e). G21) then ae) 4 + Zane =a (3.22)66 R normalization group 3.2 where a;(A) = Ab,(A). Differentiating both sides of this equution, we get dos day || dd day | wore ees ]ut =[1 ln £] eas (3.23) Equation (3.20) becomes wea, “et 8 (3.24) dail daz | ‘|- oy[i+ St Ra Assume that the 6(A) is a tinite series in 6, BOA) = [fy + Bie + Bas? +--+ Bus]. (3.25) By identifying powers of on both sides eqn (3.24), we see that the (A) senes ‘must terminate after the first power of ¢ (By = O for k > 1: \ BOA) = fo + Bis. (3.26) From eqn (3.24) we then get A=n-a 3.27) and fot hittena of foot, 32m Thus the f-function is given by d day BO) = dea FAS a HAE ws 80. 8.29) Using ay = Abi, we get fay =v (& ) : (3.30) In fact, eqn (3.24) ulso relates different powers of (1/e) in the expansion of Zor AZ), From the coetlicient of (1/6)" we have d d day\ dg dena mena = Boe + BE =(-a+ait) 4 oiast a ah ah o dys dday\ ay mings = (-a + aStt a aan ‘Thus ull the eocfticients a, with n > I cun be determined from a1. by repeated use of eyn 3.31).33. Bsfunction for QFD 67 f-function for QED ‘The photon propagator is given by 2 ?, =5 1. 3.3 1 Dwl) = Tage (3.32) where 2 3 2 u = = = 3. Teme = 28" B= TE G33) and 104) (Rv = Gude) = MvlQ) (3.34) is the vacuum polarization tensor, Define the running electric charge as 2 : 3 ——, 3.35) OTR) G39) where #(q?) = 2(g?) — (0) is finite, Show that in one-loop, we hive. Ap) = "ub + hint (3.36) for the case of u}. 1 > m?. If we define de Aled = Has (3.37) then we will huve Meat 0). (3.38) 129? Solutlon to Problem 3.3 ‘Vacuum polarization in QED ix calculated in Problem 2.6, and is given by X07) + [40 a(l—a) rien |, ta gat aA = gha{ (L+mar-y) Ct -6f daa() —aytn |, ng@)= al + ow}. 3.39)68 Ren smatization group 33 ‘Taking 4. = m, the subtracted quantity becomes. Hq?) = mq?) — 0) a7 ‘m2 — gat —a) =o - megett=a)) | ’ i f daw! a] - ] (3.40) Write 2 202) = £ xe —Rerig? 8 reer ad [1 -Rew(q?s]. B41 For the case of |q?| 3 m?, we have 2 : fe ie (+4) (3.42) 2 2 2442) met (m4 eg?) xe [! +h in( = )}: 04) Thus, us (—g?) increuses, e2(g?) also increases. Different subtraction schemes From the vacuum polarization 7(g?) given in eqn (3.39) we can illustrate the difference of different renormalization schemes, ‘Momentum subtraction scheme, In this scheme, we make a subtraction ut q? = —M?, then 2 1 a (qM) = sf daa(t -aoto[ (3.44) Suppose m? >> |q?] and Mf, then 2 2 2 gM ne (@.m) + 50 (5.2) +0. (3.45) ‘This means that a heavy fermion will decouple in the vacuum polarization at energies much smaller than the heavy fermion mass. This property will enable us {0 ignore all the unknown particles which are much heavier than the present energies. MS scheme. Here we subtract out the pole at ¢ = 0 and some constants, -4 f' mi? = Pat — a) M802) = 2% ga mig) =f dave oot cae 7 3.46) In the limit m? >> [q?] we get Sow (3) G47) non-zero, Thus in this scheme. the heavy particles do not decouple at low .. One way to remove the effect of the heavy particles is to integrate out34. Behaviour of ® near 1simple fise | point 69 the heavy fields in the Lagrangian and work with the effective Lagrangian without the heavy particles. ehaviour of % near a simple fixed point Derive the ultraviolet behaviour of @(1) in the case that the £-fuuction is given by B(g) = g(a? = °) (3.48) with a being a known constant, This example the running coupling near a simple fixed point. lustrates the typical behaviour of Solutlon to Probtem 3.4 To unalyse the asymptotic behaviour, we plot B(x) vs. R. rr) Pu. 3.1, The initial condition for the ru ing coupling constant ix RO | oat =O. ‘Then it is clear from this plot that Ra if go>O RO >a if gu <0. ‘This can be veritied by more explicit calculation as given below, (3.49) (3.50)0 Renormalization group 35 or (#0) tan = 3.51 Bday ** 7) so that a a Bs Tae oth Ae # (3.52) Taking the square root, we get ta i= tay (3.53) ‘To choose the sign, we need to go back to the initial condition that § = go att = 0. For the case gy > 0, we take the positive sign a nee ‘so that at ¢ = 0, a ao =A © Caja0) In this ease, # > a, a8 1 > 00, For the ease go < 0, we need to choose the other sign & = Ro i=——_. (3.54) ' (1 Aen2)7 ‘Then we have # + —a as ¢ + 00, unning coupling near a general fixed point point g = go, show that B(x) bas a simple zero: A(g) = —b(g — a) with b > 0, then the approach Of g(1) to gy as ¢ + 60 ig exponential in (b) if B(g) has a double or higher zero: 6(g) = —b(g — a)" with b > 0 and ‘n> |, then the approach of (¢) to go as f —* 00 is some inverse power in t. Solutlon to Problem 3.5 (a) Simple zero: B(g) = —b(g — a) From the renormatization group equation a =A®. % { dt, (3.55)3,6 One-loop renormaiization-group equation in massless dp theory 1 then with the initial condition of g + go ut ¢ = 0, we have (3.56) with b > 0, B= at (go-aye™, G57) ‘Then, % —> a exponentially in the asymptotic 1 + oo limi (b) Double zero or higher: A(g) = —b(g - a)", 1 > I. The same calculation yields dt —- =ht oF ie ~ fm '? Goh la = arnt Roa I (n= | Wasa + | enttet Ge inoe| One-loop renormalization-group equation in massless Ad‘ theory In the renormalization of the massless A¢* theory, we can momentum subtract at = to avoid infrared singularities. In this case the renormulization-group ‘equation takes the form a [oaks +05 avo] Tre preenpd 28 B59) Verify explicitly, the one-loop result for the four-point funetion P%? (pi, pa pr) sutisfies this renormalization-group equation, Solution to Problem 3.6 From CL-eqn.(2.31), the four-point function in one-loop is of the form Pst u) = iby + MGs) + PW) + Pw) 3.60) where P(p?) in the dimensional regularization scheme is given by |Cf. CL-eqn 121) rip) = 8 (2 f dattnadt In(—p?) 361 coh = sae [Fe ff aatinac —a)| = In(—p’ | G61) Suppose we nuke a subtraction at some space-like momentum p? = —M?, Then we have 2 Fp) = Pp) — (M2) = pa In (75 35 ) (3.62)n Renormalization group 3.7 where the dependence on M is rather simple (compared with the 4? dependence in the massive theory), The renormalized Green's function being POs. tw) = 1A + Pes) + PO) + Pad (3.63) we get a Bing 3ia? Mag Thiet) = ae jena! oa), (3.64) Also we have AOE, “= = =) - i+ 00) (3.65) where we have used CL-eqn (3.47) BA) = (3) +000), (3.66) Therefore, from CL-eqn (3.48) that y(A) = O(A2), we see that 8"%s. 1.0) satisfies the renormalization-group cquation to order 4?. 7 B-function for the Yukawa coupling Uhe Lagrangian for the Yukawa interaction is given by : 2 1 2 Lm ily", — my + foe +350,07- Se. (3.67) ‘Compute the Callan-Symanzik A-function for the coupling constant f. Solutlon to Problem 3.7 It is convenient to set all the masses to zer0. (Vertex correction 8 rs (ify ae 7 ya dk da ’ ith a? = —a(1 - ap? =f Jum f ea with a? (lap? (3,68)3.7. Pefunction for the Yukawa coupling n where we have combined the denominators by using the Feynman parameter and have shifted the integration variable k > k + ep. The divergent partis then atk 1 Gay Bay G9) so that at Tet = (m4 at “). (3.70) The vertex renormalization constant is then 2 Z, =14 Sanat GB.71) Recall that the A-function ix given by a; | B= Scammer where Zy = Z7'ZyZ}, (3.72) Here 74 and 7y are the wavefunction renormalization constants for scalar and fermion ficlds. Thus the contribution coming from Z, is 3 Ai y= ao. G73) (i) Fermion self-energy Ly(p) = (-if)? (3.74) Combine the denominators in the woul way ! da zB R 3.75) with A = (k= ap)? —a? and a? = -o(1 — a) p?, Shift the integration variable & = k-+ ap; the numerator becomes (1 - a) 6 — &. Then we have 1 (as - 41 if oe Ba Typ) =f “fs 2 i Ae =P f'aa-er gy (4+ (3.76) and the wave function renormalization constant is oe In A? +. G77) and its contribution to the £-function is Zy=t Pref zy ee Ze 3.78 ainay2* ™ Tent cae Ice there are two such diagrams in the vertex. this contribution should be by two.” Renormaitzation group 3,7 (iil) Scalar self-energy Lolp) = (if)? (-) [& earr( dk rik - #91 =O? | Gt ph 7 ‘The numerator is, in the dimensional regularization, N=TrHE(S —#)) ed (pk). (2.80) The denominator is, moor: [ot with @=-a(l-ayp? 381) Bo-kt hy (BaP where the shift k — & + ap has been made. With the shift the numerator is then WN mdl(k + ap) — p(k +ap)] > dik? = a — ap] = dk? + 0), 3.82) The self-energy is Lo(p) = Hae +a). (3.83) Using the formulae Cd iy" Pwd fa) (14? Sea ~ Uae rer (at) ed atk woe SS pee 4) ans Gm at ~ Gm Ty easy we get atk (+a?) i (2 ay =a ~ ny i Gn als YS 3 NES a SN 1 a ois LY + => ~ 1 1s = (3.86) ‘The self-energy is then (3.87) Papa fh , Maen3.8. Solving the renormalization-group equation by Coleman's method 18 ‘The divergent part, which is relevant to the wavefunction renormalization, is then, with d > 4.and P(2 —d/2) + (2/(4 -d)), sit / 2 mip? sf? LolPaw = Era? a)- ser? (3.88) where we have used the correspondence (2/(4 ~d)) — In A? (see CL-p. 56). The wave function renormalization constant is atta Zomlt eatin 3.89) and its contribution to the B-function is p =—,2. .! Ama 3.9) The total contribution to the £-function i = fe we B= (B+ 2Pr +B) =o 24axt 3+2)= iar B91) ix the result given in CL-eqn (10.16). Solving the renormalization-group equation by Coleman's method Consider a one-dimensional fluid with velocity u(x) and in the fluid there are bacteria (see Coleman 1985). Let p(7..x) and g(x) be the density and the growth rate of the bacteria, respectively. (a) Show that the density of the bacteria p(t, x) satisfies the differential equation a @ lz tue - «eo pt.x) = 0. 3.92) (b) The position of a fluid element is described by ¥ = (1.x) with the initial condition ¥(0, x) = x. Namely, the fluid clement which was at x at ¢ = 0 is now ‘at X at time ¢. Clearly X(7, x) satisfies the differential equation Signy = 00. (3.93) a Show that if p(0. x) = p(x). then at later time p(t, x) is given by 7 P(t x) = pol X(t, x)) exp Lf acre’. 2n] . G.94) Solution to Problem 3.8 (a) The term due to the growth mte g(x) is self-evident, We will concentrate on the second term which is due to the motion of the fluid.16 Renormalization group 38 Consider a fluid clement f with length dx located at x. The bacteria in this fluid element is just p(t, x) dx. Ata later time, ¢ + Af, this fluid clement is replaced by those which were located at (x — vAr) at time f. Thus the rate of change in the bacteria density in f is lo(t.x) = alt.x = vAN] ae ar cae Ok where we have made the approximation p(t, x — var) = p(t..x) — vap/ax ar. ‘This gives the second term in the differential equation. (b) Integrating eqn (3.93) for Z, we get (3.95) dy ay 3.96) [ u(y) 690) We can differentiate this equation with respect to x to get 1 or a4 ao, 3: vax ~ vay =” ed Then for any function f(x) we can show that ie (at 7 fttan = SOF = s/w, wo fet = woe = fe). G98) Fane ox : Combining these we get a Lae [3 - wore] SEU, x) =O, 3.99) Or changing t + —1. we get a a] ns [3 + wre] E(t) m0. (3.100) ‘Then it is straightforward to verify that the solution is of the form AUX) = polE(—t,x)) exp LL argccr.2p). G10) lo Remark, ‘The gencralizution to a higher dimension is simply a [Ft me... Bay RO | elt. x1. 3.102) Define a. A — Tite Rte tad = UR with 80.1 te) ee (3.103) ‘The solution is then PL Xn) = pol F(t x1 amen] f dr x, (3.104)3,9. Anomalous dimensions for ¢ posite operators 7 Anomalous dimensions for composite operators In the A¢* theory, compute the anomalous dimensions for the composite operators, @? and 6°, in the one-loop approximation, Solution to Problem 3.9 () Anomatous dimension of ¢? As was described in CL-S Joop divergent graph involving ¢ form a ieih\ (dtl l i PBC: sma) = ( +) ar lFoal lapse} Ge ‘This has exactly the same structure as the function I"(p?), given in CL-cqn (2.70), which appears in the four-point function, Taking over the result, we have on 2,4, the only one= n the two-point function 1G), and is of the rp; mA, A? dein? — a) —@)p2] 4 Ble: Pu pad nap, deine? — at — aap] + 32m? (3.106) and Ze = 1413000) = @.107) The unomalous dimension ix then a a Yom Tina Oe = Tent @.108) () Anomalous dimension of ¢° ‘The one-loop divergent graphs are all of the type shown in Fig. 3.3 with altogether (5) distinctive diagrams. Pa 34 This agnin can be expressed in tenns of I"(p?), Taking into account the contbina- lorics we get ISA A? Zo m1 + 1ST 30,00) = 1 ml - (3.109) The anomalous dimension ix then Yo= (3.110) SA 163"Group theory and the quark model Unitary and hermitian matrices Show the following relationships between the unitary and hermitian matrices: (a) Any m xn unitary matrix U'U = | can be written as U = exptiH) a) where H is hermitian, H* = (b) det Um | implies that #7 is uraceless, Remark. This result means that 1 x 1 unitary matrices with unit determinant can be generated by # x 1 traceless hermitian matrices. Solutlon to Problem 4.1 (a) The matrix U can always be dingonalized by some unitary matrix V VUV' = Us (4.2) where Ug is a diagonal matrix satisfying the unitarity condition UyU} = 1. This implies that the each of the diagonal clements can be expressed as a complex number with unit magnitude‘, Us . (43) em where as are real. It is then straightforward to sec the equality Ug =, where H, is a real diagonal matrix: Hy = diag(a, a... dn). We then have UsVtugv = Vt My well 4) with H = V'H,V. Because H; is real and diagonal, the matrix H is hermitian: HY = (vii)! = VtIV eH. 3) (b) From the matrix identity e” = det(e4), we have for U = eM” IH wm det (0) = det U. (4.6) ‘Thus det U = | implies that TrH = 0.4.3. Reality of SU(2) representations pn sU(#) matrices ‘The n x 1 unitary matrices with unit determinant form the SU(n) group. {u) Show that it has n? — 1 independent group parameters, (b) Show that the maximum number of mutually commuting matrices in an SU(m) group is (= 1), (This is the rank of the group.) Solution to Problem 4.2 (u) To count the number of independent group parameters, it is easier to do so through the generator matrix. From the previous problem, we have U =, where H is an mx m traceless hermitian matrix. For a general hermitian matrix, the diagonal clements must be real, Hi = H,3. Because of the truceless condition, this corresponds to (n ~ 1) independent parameters. There are altogether (n? — 1) off-diagonal elements and thus (n? — 2) independent parameters because each coinplex element corresponds to two real parameters, yet this factor of two is, caneelled by the hermitian conditions H,y = Hj; Consequently, we have a total of (1 — 1 +n? — 1) = (n? — 1) independent parameters. (b) Fromthe discussion in Part (a) we already know thatthere are n— | independent diagonal SU(n) matrices, which obviously must be mutually commutative. On the other hand, if there were more than n1— I mutually commuting matrices, they could all be diagonalized simultancously, thus yielding more than n — 1 independent diagonal matrices. This is impossible for nx 1 tracelexs hermitiun generating, matrices, Reality of SU(2) representations This problem illustrates the special property of the SU(2) representations, their being equivalent to their complex conjugate representations, (a) Forevery 2 x 2 unitary matrix U with unit determinant, show that there exists 4 matrix $ which connects U to its complex conjugate matrix U* through the similarity transformation SUS = U*. 2) (b) Suppose yr and 2 are the bases for the spin-} representation of SU(2) having cigenvalues of 4 for the diugonal generator 73, Twh = fyi and Ta =—}ya. (4.8) calculate the eigenvalues of 7; operating on Wf and yj. respectively. Solution to Problem 4.3 (a) We will prove this by explicit construction. Problem 4.1 taught us that the unitary matrix U can be expressed in terms of its generating matrix U = exp!80 Group theory and the quark model 4.3 ‘Thus the matrix S, if it exists, must have the property of SU'HS = -H™ (49) so that S“'US = S~'(expilf)S = U* = exp(-iH"). The generating matrix H; being u 2 x 2 traceless hermitian matrix, can be expanded in terms of the Pauli matrices MH Saya, +440, +4 (4.10) with real coefticients of expansion as. Since a1 and ay are real, 2 imaginary, we have H® way — 4202 + ayy. 1b Equation (4.9) can be translated into relations between Sand Pauli matrices: ‘aS = -a1, S~!028 = op, and S~'oyS = —o5, Namely, the matrix S$ must commute with 2, und anticommute with « and 03, This can be satisfied with Saco; (4.12) where *is some urbitrury constant, Ifwe choose c= 1, the matrix is unitary and hermitian; for ¢ = i, S is real. (b) ‘The statement ‘y and Ya are the buses for the spin-4 representation of SU(2)* means that under an SU(2) transformation (i = 1, 2) Ws vim Uy; with U =explie-o), (4.13) lo matrix notation, this is y” = Uy. The complex conjugate equation is then Vr aUY a(SUSY of (SY) SUSY). (4.14) This means that $y has the sume transforination properties as y. Ex ion, we have (2) ae To say that it has the same transformation properties ax ws ) 4.16) zat “ ‘means that, For exaunple, #)A(2 2) (4 1 (4) = (9 ta) (44): am Namely, the eigenvalues of the 7) generators are AWD =A) =} (YD) = tha) 4.18)44. Antidentity for unitary matrices 8) Remark, This shows that the 7 = J representation is equivalent to its complex conjugate representation, We say that it is a rea representation. This property ‘ean be extended to all other representations of the SU(2) group. because all other representations cun be obtained from the 7 = } sepresentution by tensor product. Part (b) shows that the matrix S transforms uny real diagonal matrix, e.g. 03. into the negative of itself. In other words, $ will transform any eigenvalue to its negative. Thus the existence of such a matrix requires that the eigenvalues of the hermitian-generating matrix occur in pairs of the form tay, ka, ... (or are zero). 11 is then clear that for groups of SU(m) with a > 3, such a matrix S cunnot exist as eigenvalues of higher-rank special unitary groups do not have such a special pairwise structure. An identity for unitary matrices (a) Show that if A and B are two n x n matrices, we have the Baker-Hausdorft relation 2 ef Be'* = B+iA. BI + FlA 1A. Bite TATA... TAL Be bee (4.19) (b) Show that the matrix B is invariant (up to a phase) under the transformations generated by the matrix A. if these two matrices satisfy the commutation relation of [A Bl = Solutlon to Problem 4.4 (a) ‘The matrix J, defined ax J(A) me Be~4, being a function of some real Parameter A, can be differentiated to yield: ds _ any ana, ad A = iA, Blew => =] mit, By IC, a leo BI ay aoa CL 2 2 ae TOMPIATA, Bile op 5) = PIAA. BIL PC: PY 8 ain nA ay CL " a ae ne TA. Cale? > TF hoo TA, Cy-1] Cu Expand J (A) in a Taylor series: Jas x v1) x Sire, = (4.20) SO lag nl em ' = B.C = IA. Rh and C, = desired result A, Cua]: Setting & = 1, we have the 1A Ben i e Bel mB FHA M+ SIALA BIA 42182 Group theory and the quark model 4$ (b) To show that ‘the matrix 8 is invariant (up to a phase) under transformations generated by matrix A* means to show that (aA pe“'0A = B (4.22) for un arbitrary real parameter c. But from Part (a) we have already shown that A Belt = i (4.23) 2h where Co = B,C, = [A.B and C, = [A.C,-1]. For the case at hand of (A. B] = B we have C, = B forall n= 0,1... 2 Bett = BY (4.24) ra ‘This is the claimed result. An identity for SU(2) matrices Prove the identity for 2 x 2 unitary matrices generated by Paull matrices o = (1, O34 09): ' explir 0) = cour-+(@ + o)sinr (4.25) where r = |r} is the magnitude of the vector r and # = r/r is the unit vector. Solution to Problem 4.5 ‘We will first derive a useful identity for Pauli matrices. Consider the multi of two matrices: (A+ MB - 0) = (0,04)A, By = Hoa; +0; 0) + (aa, 0) NAB, = $(loy. a) + loi. DAB, = $(25, + ZieyondAs By (4.26) where we have used the basic commutation relations satistied by the Pauli matrices: lay o)) = 2ieyeoe and fo, ay} 2s (4.27) ‘Thus we have the identity (A-@)(B. 0) = A-B+ia (A xB) (4.28) Set A = B= r, we get (t+ 0)? =r? 4io-(r xr) mw rand (Fo) =r7(rog) = r+ o). It is then straightforward to see that ant "(Bo (4.29) (ray er and (ro)!46 SU(3) algebra in terms of quack fields 8 with = 1.2,... The desired identity for the unitary matrix then follows as expir-oy= oe 5 ig e ta hate a =cosr + (+) sine. (4.30) Remark. This relation holds only for 2 x 2 unitary matrices and does not hold for highcr-dimensional cases, where anticommutation relations are much more complicated than just the Kronecker delta. U(3) algebra in terms of quark fields (a) Given a set of composite quark field operators ' - farerzaod's 4.3) where the quark field operators ‘qulxV g(x) qulx) (4.32) 42), commutation relations Gal) gn) = ba B(x — ys (4.33) 3 oye und where d/, with i = 1, 2, (4.34) show that { F'), if assumed to be time-independent, generate the Lic algebra SU(3): (FL FA sacl re, (4.35) (b) Calculate the commutators | W2, W:'] for the non-hermitian generators: w! f gnqalx) dx. (4.36) Show that W4 is just the ikoxpin ruixing operator. respectively, the U-spin and V-spin raising operators, similarly, W? and Wj are,a4 Group th ary and the quark modet 44 Solutlon to Problem 4.6 (w) The proof can be obtained by applying to the commutator [F, F/) the iden- tity of [AB, CD] = -AC{D, B) + ALB. C)D — CfA, DIB +{C. A}DB, (4.37) which for the present case has {A.C} = {8, D) =0, - aw (P= f ded'y ao (F Gh). aon(S 4c) 2) 2a i J = [ese'yce-vfaten (3) bn a) guy) 2 ab 2 od =q! x) ) } aon at 2 ee iy Px qt [ Jan Tian (4.38) where, because F's are assumed to be time-independent, we have chosen xo = Yo for convenience, and applied the equal-time anticommutator of the quurk field operutors. (b) Again from the identity eqn (4.37) and the quark field nticommutation rela- tions, we have [wh we] = f Pad y [ghexrgets. ahivaet)] = f Ba dy Mex — yy [aftertlaeds — aor ats] = tw we, (4.39) If we write (g1.g2-qx) a (usd). the non-hermitian operator Wy =f dorards = felerduna's (4.40) is shown (o be an operator which transforms a d-quark to au. Clearly WS E : isospin raising operator, Similarly, we haves “> d and ¢ > u. Remark. In this notation the third component of the isospin generator Ty tukes the form of ty m4 fore wd'dyd's i fata qd’ = fw! Wy 441) and the hypercharge in the form of yet fou dtd —2stsyd'x = HWE + WR-2WY, (4.42)4.7 Combining two spin-} states 85 Combining two spin-} states Consider a doublet Y= (Wr. ¥2) of the SU(2) group. Show that the composite operators Se yy and Vis yt ry. r= (ty 12.75) being the usual Pauli tmutrices, transform as a scalar and a vector respectively. Also, demonstrate the vectorial transformation property of V in several ways: (a) fora general infinitesimal rotation, (b) fora finite rotation around the 3-axis, (©) fora general finite rotation, Sojutlon to Problem 4.7 Under the SU(2) transformation, we have vo yseety, plo ptayide (4.43) where a = (or, a2, 9) ure the three arbitrary real parumeters, It is clearthat yt y is an invariant under SU(2) transformation, Saypty sytee! amy syty =. (4.44) (a) To demonstrate the vectorial transformation property of V under a general infinitesimal rotation Ww ya (l ia: 2/2)y. . vs (4.45) yl pt = wt ia: 2/2). We note that the transformed composite operator can be written Vie p try’ zetia: 1/21 ia: r/2y su'(rsi (S ‘] +01) y. (4.46) But the commutation relation for Pauli matrices yields Fn] =e,/2 1] fare puts | say |S. | mien 7 Ryser lst 1 (rx we. (4.47) ‘Thus we have demonstrated the vectorial nature of V under the infinitesimal SU(2) transformation Vew(r-rxmyev Vxa. (4.48) (h) To demonstrate the vectorial transformation property of V under a finite rote tion around the 3-axis: WaRY WT my with Rae ion, (4.49)86 Group theary an I the quark model 4.7 the transformed V operator can be written V ey ry SUR Ry = plan? zeny, (4.50) Applying the formula of eqn (4.19) 2 tre ti [2S Sf (ao oe Aree +i[S2. e] 45 [Se [Sel] + «43d we clearly see that, because {(13/2). ts] == 0, and thus R'yR = 1. the third ‘component of V is unchanged under a rotation around the 3-axis, For the other two components weneed to calculate [(t3/2), 2]. This can be considerubly simp! if we work with the combinations ry = 1, + it, which obcy the commutation relation of [F. ] =n. (4.52) In panicular, we have (e1- 3. nf] =y (4.53) ) ( Heian) + Himsa? +--+) =n, (455) ind thus, Reka ry ( +a) + piles? 4+) mye, 434) RR It then follows that Rig R= 5 (tee + re) = cosayt, — sinavny (4.56) and similarly sin sty + cosestz. (4.57) 1 RR = 3 (t+ el 4 etlas! Consequently, the three components of V have the following transformation Property under a finite trunsformation around the 3-axis: Vi = cosas; — sina V2, Vj = sin ayVi + cosa Va, Va Vy. (4.58) (c) For the case of arbitrary SU(2) transformation, we have Viewty =y'u'ruy where U meter, (4.59) vv ere) eS) Te48 The SU(2) adjoint representation 87 The basic commutator can be calculated as follows: (FS). »] =a, [# x] ayiewin = (0 Out (4.60) where we have recognized the spin-1 representation of the rotation operator (pr = He (4.61) or the double commutator. (EG ae ole] S (oe - Qu (oe Oyty me (-0- 0,7) (4.62) and so on. Thus U'nU = [! + (ia t)+ z (iat)? + a ye (et. (4.63) or Vy Vim (OD Mee (4.64) The SU(2) adjoint representation (a) Suppose ¢ transforms as a vector under SU(2) as discussed in eqn (4.64): Oy OO dee (4.05) ‘Show that the transformation law for the 2 x 2 matrix detined by & = r-¢ is given by b— o'eu'éu with Uwe" (4.66) (b) Suppose & is a 2 x 2 hermitian traceless matrix which transforms as $—Pevtiy wih Uae, (4.67) ‘Show that £' is also hermitian traceless, und with det £’ = det £. Since £ and © are hermitian and traceless, they can be expanded in terms of Pauli matrices Lar-@ ond Lar ¢. (4.68) Show that and ¢ are related by a rotation. (€) Suppose we have the nucleon in the isodaublet representation NV = (p11), and the pion in the isotriplet representation x = (7x. 72. #3) with m3 = 2°, (my = tg) V2 = 1, and (4 + imz)/ V2 =~, construct the SU(2) invariant pion-nucleon NN coupling.8B Group theory and the quark model 4.8 Solution to Problem 4.8 (a) We will show that &’ = U*OU follows from the transformation eqn (4.65): Oty = (Cadet (4.69) sali+ciens tient] a 2 ut which can be written, according to eqn (4.63), as, & = gU'nU =U ou, (4.70) Remark. This problem shows thut there are two alternative ways to describe the trunsformation of the vector representation (more generally, the adjoint represen- tation): as in eqn (4.64) or as in eqn (4.66). () To show that . is hermitian if £ is hermitiun: = (U'Euy' = U'éu sutéu = 8, (a7) Vo show that £’ is traceless if ix traceless: wf! srrutiu siruu't srk 30. (4.72) To show that det £7 = det £: det £ me det UTEU = det UUTE = det £. (4.73) Expanding £' and £ in terms of Pauli mutrices: Large ( Mere oe). (4.74) it is easy to calculate their determinants: det £ = - (4) + 92 + ¢3). (4.75) ‘Thus the above result of det £' = det £ implies that the transformation ¢ —> ¢" leaves the length || unchanged. This must be a rotation. (c) From (a) and (b) we see that the 2 x 2 hermitian truceless matrix Ae rem, formed from an SU(2) vector x, transforms by the similarity transformation: fico frsutfu wih v= “9, (4.76) In this form, it is easy to see that the product N' TN, where N is an SU(2) doublet. is invariant under SU(2) transformations. This suggests the invariant pion-nucleon 7 NN coupling to be : 0 + Lanne gh ro = gp, a(R Vin’ ) (?) -7 = a[(ip -finyn® + Van + pny). (a. and thus the relations among coupling constants are I Rinnt = Faint = Felon = R- (4.78)49 Couplings of SU(2) vector representations 89 Remark. The relation gine+ = Ripn- follows from the hermiticity of the Lagrangian density (or charge conjugation). Couplings of SU(2) vector representations (a) The p vector meson has isospin | (it has three charge states: p*, 6%. p7). Construct the SU(2) invariant x7 coupling. (b) ‘The w vector meson has isospin 0. Construct the SU(2) invariunt wor coupling. Solution to Problem 4.9 (a) Since p has spin | and isospin 1, we can represent the p-ficlds as p,(x) and define a 2 x 2 matrix B, = + p,. As one has seen in Problems 4.8 and 4.9, it trinsforms under SU(2) as B, — PL wut hu (4.79) just us the pion matrix Fl —> AY’ = UA, where U is an arbitrary 2 x 2 SU(2) tmatrix. This suggests the invariant coupling to be Lon = gir by’? FF) =ar( pe view") ( n° ven) ( rn? vn") v20# =p J\ am au J \ Jin =a! = ¢ [20 (ant — ann) + 20% (,n-n” — dan) +26 (a,ntn® - ayn°n*)]. (4.80) ‘This implies, for example, the equality of decay rates Po a rt) =Pot ot =Pip > on), 481 Remark 1. The decay p° + 7°r® is forbidden because the (17°) system can only have even orbital angulur momentum because of the Bose stutisties. Hence ‘angular momentum conservation will forbid this decay. Note: the same argument can be applied to the vector gnuge boson Z to forbid the decay into two identical Higgs (scalar or pseudoscalar) bosons. Remark 2. The other possible coupling #r(P,, 12") is not independent of the One just considered. This can be seen by applying the Pauli matrix identity (Cr ACr B) = (A. By ire (A x BY (4.82) which implies that r(Bantnin) =i p,(0*x x 7), (4.83) tr(Aefonht) = ip, (an xm. (4.84)90 Group theory and the quark model 4.10 (b) The SU(2) invariant Lorentz scalar combination out of the p, 77, and meson fields can be constructed as Lome = gtrP, Ej" = 2g (ofr + pint + pln") am. (4.85) Isospin breaking effects Exact SU(2) synunetry implies the degeneracy for particles in the same irreducible representation of SU(2). But the SU(2) isospin symmetry is broken in nature by electromagnetism as well as by up-and-down quark mass difference. The first-order electromagnetic breaking, involving the ernission and absorption of a photon (and thus the electromagnetic charge openitor Q acting twice), contuins un isospin- changing A/ = | piece, as well as a AT = 2 piece, On the other hund, the quark maxs-difference: mye + md = (iu — dd) (4.86) contributes only a 1 = 1 breaking, ws the last term transforms ax the third component of the isospin generator 1. Thus the strong interaction Hamiltonian can be written as H = Heo + Hy + Hey (4.87) where Hey ix SU(2) invariant, is the O/ = 2 electromagnetic breaking term, and the A/ = | piece 74;,, contains both the clectromugnetic and up-down mass-diflerence breakings. In this problem you are asked to calculate the first- order mass shifts due to #1’ by using the Wigner-Eckant theorem for the following isomultiplets: (a) P=} (pom, (b) £= 1; (E+, DE), (c) fm fs (At, A+, 0%, An). Solution to Problem 4.10 According to the Wigner-Eckart theorem, the matrix elements of a tensor operator O}f having isospin 7 and third component value M. have the simple structure of (POF Lb) = (08) TM 8 YT Orys) (4.88) where the first factor on the right-hand side is the Clebsch~Gordon coefficient and the second factor is the reduced matrix element, which ix independent of #3. H- and M. For this problem, the operator O = Hj, and Miz, which transforms as the (T = 1, M = 0) and (T = 2, M = 0), respectively.4.10. Isospin breaking effects 91 (a) {= $ multiplet (p. 1). ‘The first-order mass shift duc to 71, can be evaluated. by the Wigner-Eckant theorem as 5p = {0 |My] #) = {2-4 Mol be) (hg 1.05 f. 7)8 my =, (4.89) and 8 tn = (n [Hey] = (h. —$ [Heal fs) (fe $[10f, -$)5%my = ffm 4.90) where my is the fy-independent reduced matrix element. Exactly the sume culculation shows that the AJ = 2shifts 82m, = 82m, = asthe corresponding Clebsch-Gordon coefficients vanish (because an = 2 oper- ator cannot connect two f = 1/2 states), In this way we find that the proton and neutron mass shifts Smt p= 5a p.. have the same magnitude but are opposite in sign: Simp = —Bry. 9D Remark 1. We can upply this result to any other / = $ multiplets. For example, dmg = —bmg-. bigs = Bug «ele. (4.92) Remark 2. Alternatively, we can write down an effective mass term in the | agrangian, which contains an operator having an isospin value of (T= 1, M = 0). For the isodoublets this cun be represented by a 2 x 2 matrix, ts. The effective mass term for the nucleon can then be written as Lamy = NEmy tN = bmn p. 7) ¢ 3 ) (?) = bmy (Bp — fin), (4.9%) which yields bmp = —bm, = Sy. (4.94) (b) 7 = 1 multiplet (L*, £°, £-) The Wigner-Eckan theorem yields 8M mgs = (LP [Ay] L*) = (1. +1 [Peo] bs +1) Vem: (4.95) 8 mge = (E° [Pe] £°) = (1.0 [Hey] 1.0) = (1.0110; 1.08 me = 0 (4.96) mL FUL OF 1 $18 my92 Group theory and the quark mudel 4.10 }) Demy = — Sams: (497) 5% mg- = {E> [Hy] E7) = (11 [Mn be =({1-Ilh and 82 mgs = (EF [Mey| E*) = (1.4112, 0:1, +198 ms = Roms — (4.98) 8% mye = (E° [Ajay] E°) = (1. 012.05 1.08% mp = - 7% ms (4.99) 8% mg- = (EX | Hey] E7) = 1. 112,05 1, 198mg, =f feP me. (4.100) Combining these results and using the notation my = /48t™ms and my = [8 mz, we have my no + my + nay mye = my + 0 — 2m, my- mg = my +m. (4.101) ‘The 1 = 1 und = 2 muss splittings can then be isolated: my = }(mg» — mr), (4.102) mg = LOmgs + ng- — Inge). (4.103) While nty contains both the electromagnetic and up-down mass difference effects, ‘mz is purely electromagnetic in origin, Remurk. The sume analysis holds for the isotriplet pions. In particular, we have, besides 5Ymgo = 0. the result 8m = 5PM (4.104) But 2r* is the antipurticle of x~ and should have the same mass. Hence mgr =Smy- (4.105) ‘The only way to reconcile these two eqns (4.104) and (4.105) isto have the reduced matrix element 8m, = 0. (The Wigner-Eckart theorem does not by itself give any information about the reduced matrix clement.) The same result can be seen from writing out the mass term in the Lagrangii Loti) = AM my or (Fett) a x Jint\/1 0 x Jint nate ANS VK = A ny (1 — nt — 1 4 2ntA7) =O. (4.106) ‘Thus the pion mass differences (and for p mesons also) are entirely due to the 1 = 2 electromagnetic corrections.4.11 Spin wave function of three quarks ” (c) 1 = 3/2 multiplet (A**, A+, A", A>) ‘the Wignet-Eckart theorem yields Singer = (EAR LOB 49) mg = FSP mg Inge = (Pf 105 8.44) 8% nn = hs mg I mge = (J, $1.0: $8 mg = yf dO BP mg = (F.—F/1.05 fF) = Siem, (4.107) and BO mngee = (E820: 8 +8) 8%me wf P3% ma 5 mgs = (J. 44] 2.0% §, 44) 8% my = fang BP ungo ={¥,—$] 2,0: $4) 5P mtg = yf fOPM mg 8 mg. me (J. —§]2.05 9) 8%ma = f}8%ms. (4.108) Combining these results we have Mars = mg + 3m, + m2 Mat = mo +m, — m2, Myo = Mg — my — mM, Sry + ma. (4.109) Mo = mo Besides allowing us to isolate the / Also iinply a mass relation of and / = 2.maxs splittings, these equations Maer — Mg- = mas — Mtg). (4.110) ply reflects the absence of an / = 3 piece in the symmetry-breuking. lonian, Spin wave function of three quarks Asan exercise in Clebsch-Gordon coefficient calculation, construct the spin states, Of baryons, which are composed of three spin-} quarks. Solution to Problem 4.11 The possible spin states for two quarks are Sjz = 0,1, where Siz = S; + $2. As. discussed in the text, the Sig me I states [Si2. Siz.) are HD) = aap 11.0) = Jp(aufa + Broa) 1,1) = App (4.1 By4 Group theory anu the quark model 4.11 where o = |}. 4) and 6 = |}, —4) are the spin-up und spin-down states, respec- tively, Also we huve the spin-zero combinuti 10,0) = Aiaiha = Bia). (4.112) Combining the Siz = | states with Sy = }, we obtain $ = } and 4 states, where S$ =S.. +8) =S, +8, 4S. |S= 3S =H =I. Db f= cree. 4.113) To reuch the other S = ¥ states, we use the lowering operators SIS. S) = [(S + SS - S, + D]'? IS. S;- 1) (4.114) to obtain Ss. lf. §) = v3|E. 4). (4.115) On the other hand, S_ = (S12) + (Sy). Thus SAV |E §) = Sd 1D | $+) #1 IS [5 4) = VAI. O)ay + 1. DBs. 1 4116) Combining eqns (4.115). (4.116). and (4.111, we get Ped) = 3s [V2I1. 005 +11. 1] = Sylteipa + Prendos + avenps] = Jplaupras + Prozas + cro2fs]- 4.117) Similar to eqn (4.113), we have [$= §.S:=-H) = 1.1) [L. -$) = Bifabs 4.118) and using Sy we can obtain |}. 4) = Jylen Bats + Braap + Bi Pron). (4.19) The state |S = J. S, = }) must be orthogonal to |S = 3, S; = 4} in eqn ( 4.117): [be Hs = du [Hit O03 + Vit. ny] = Hl2m 02s — (Proz + orPadard. (4.120) where the subscript S signifies the symmetry property of the stute under the per- mutation of quarks 1 +» 2, Similarly, we have 1h. Hs = el2Bitrars — (ane + B102)Aa). (4.121)411 Spin wave function of three quarks 95 But |}. 23) can also be obtuined from combining the Siz = 0 and Sy = } states, Such combinations ure antisymmetric under the permutution of quarks 1 <> 2: [ped = yleuBe - Aranda [de -a}, = JyloBe — Brands. (4.122) 1 summarize, we have four S = 3 states, which are completely symmetric under any permutation of all three quarks (1, 2, 13.8) = cronas [pe 4) = dyleupaoy + Brazas + och) 1. -$) = JylenArBs + Prot + AiBreal 1}. -}) = Bibbs (4.123) and (wo S = $ states |}, 4), m xu.s, which have mixed symmetry with respect to the permutation of (1, 2, 3) but are symmetric under the permutation of 1 <> 2: [he H)s = Jgl2anaas — (Bro +01f2)0] [pHs = Fgl2r Bre — (ou Br + Pra) Pr (4.124) and (wo S = $ states |}, 4}, m xw.4+ Which have mixed symmetry with respect . Tes permutation of (1,2, 3) but are antisymmetric under the permutation of [fe Hq = Jalon Be — Brandon bs Hy = Yyler Be — Brea) By. (4.125) Remark. If we are interested in the isuspin of throe non-strange light quarks « and d (us for the nucleons and the A resonances), we can work out the corresponding isospin wave functions by the simple substitution of a > wand -> d: (u) Symmetric isospin | = § statex At =|} P=auen Ot mm |F §) = Sylrcdauy + duurauy + uiuadil AM = |. -$) = Spliced + dads + didausl A” = |f. 9) = diddy. (4.126) (b) Mixed-symnmetric (but symmetric with respect to the interchange | +> 2) Tm} states ou.st [de Hs = Jylduiaads = (ditt + wields [fs Hs = pl2hetaurs = (uid + chnadds|. (4.127)9 Group theory and the quark model 4.12 (e) Mixed-symmetric (but antisymmetiic with respect to the interchange | «> 2) T= } states dyat [ped] = qyeree — dine [bea = Spence = dinadds. (4.128) Permutation symmetry in the spin-isospin space Show that the spin and isospin combination of the mixed symmetry states dis- cussed in Problem 14.11, xa.sdu.s + Xw.ada.a is invariant under the general permutations of particles i int: Such permutation operations are repre- sented by orthogonul matrices in the 2D spaces spanned by mixed-synimetric spin ‘wave functions, x1 = Xu,s and X2 = Xw.as OF isospin wave functions, @ = @y.s and d2 = du.a. Solution to Problem 4.12 The general permutation of three indices can be denoted as 123 =(i h a) (ata )—thus a permutation of the purticle in this permutation group $3: where (1.2.3) are replaced by (ln indices (1, 2, 3), There are six clemet ma(h2 8% GER) m2? 123) , (123 123 ra= (1 3 ). Pra G 1 2): (; 2 3): Its clear that under any of the permutation operutions, the mixed-syimmetrie spin wave functions x4, and xy,4 transform into linear combinations of xw,5 and xu.a- The trivial examples ure / und Pj. By construction, we have Praxn,s = XMS0 Paka —XMAv (4.130) i.e. xu.s and Xx,q are eigenstates of Py, and 1. The operator / and Piz can be written ax in orthogonal matrices: apf) O\ (x Al O\ in ig+(4 1) (3): Peak (0 2) G)- an The more general cases can be exemplified by Fix acting on xu [f.+b), = Jyl2onc2ps — (Bier + a1 P2dars}, = [f.44), = dylorBy — Brazos. (4.132)413 Combining to fundamental repreventations 7 to yield Pasxy = Zgl2asa2fi — (Aso + a3f2)ou1 =-}u- fx. (4.133) and Paxa = Slava — Brandon = $x + fx (4.134) Thus P15 can also be represented by an orthogonal matrix Pug (3 3) (*) : (4.135) is not difficult to convince oneself thut, because the permuted states Px —> x! mutst remain orthogonal to each other, all the xix permutation elements can be represented by orthogonal matrices: PUP = PPT i, (4.136) From thix property it follows that the combinations such us Beart Reandt um 4.13) re invariaut under all the permutations of the particle indices. Remark. Since the combination xy.séu.s + Xa.aém.a is invariant under any permutution of quark indices, itis totally symmetric in the xpin and isoxpin xpace. From this we can conclude thut the nucleon wave function, which is a product of spin, isospin, and the totally antisyminetric colour wave functions, is totally antisynimetrie with respect to interchange of any of its three quarks. This is con Patible with the requirement of the (generulized) Pauli principle as the nucleon is system of fermions (quarks). Combining two fundamental representations ‘Work out the tensor products of the defining representations of SU(2) and SU(3): (u) Let y = (%!) = ()) be an isoxpin doublet with its hermitian conjugate being W! ms (Wi yg) = (u a). Find the isospin of the product yy; (where i = 1, 2). wa Ww (b) Let y= *) = (‘) be an SU(3) triplet. Decompose the product wf yy Ves, s (where jm 1, 2.3) into irreducible representations af SUC).98 Group theory anul the quark model 4.13 Solution to Problent 4.13 (a) 11 is useful to denote the complex conjugate y? by y/. Namely, ym yy From the fact thut under an SU(2) transformation, SW = Uy) =U), (4.138) where U/ m Uy and U is unitary, we have Ui Ut = Ua = aU) (4.139) or Wi Wm WUS (4.140) where Uj m (U7)*. It is clear that the combination yyy is an SU(2) invariant: WU) = WUSU Na = WU Uae = WB = Way. (4.141) ‘Thus yy, husisospin | = 0, Itis easy to see that the remaining three combinations in the product yy transform as an 1 = I triplet. We can remove the 1 = 0 combination by the following subtraction; 7) yyy — 48, a) , 4142) which has the property of 7/ = 0. The 7/ components can be explicitly written out to be: T= viva utd ~ aww sdu~nt and Ths yy — 5 (Wyn + 07 ya) =F (wv — wy) = futu—d"d) ~ Fn° (4.143) TB = Wa 4 (Wh + Va) = 4 (W'va - Wa) = —futu-dtd)~ sn°, (4.144) ‘Sometimes it is convenient to write T} as a traceless matrix a(t TRY te ( iy) We cun summarize the result in the form of a direct sum: Bx2aie3, (4,146) (4.145) where the representations are denoted by their respective dimensions. The triplet Hled the adjoint representation of SU(2).4.13. Combining two fundamental representations 9 (b) Again the SU(3) invariant trace wi is an SUC) singlet. The remaining cight components transform as the octet repre= sentation under SU(3), sulutdtdtsts (4.147) 3x whts (4.148) Following the same procedure as in (a) we cun display the adjoint representation of SUG) as Ay = yy — 48). 4.149) ‘To display the quantum numbers of various components: uld~m-, Abmdtunnt, Al=stu~ Kt, Ala d's ~ KS, (4.150) Tor the diagonal elements Al eta eta, 4.151) 3 v2 v6 where at 5 Vat i a = tutu —a"d), elu'n + d'd — 2sts), 4.152 wom eu Poe ) ) Similarly, we ean work out 2° and AW =, 4.153 5) Ve ( ) ‘These octet components can be organized as a traccless hermitiun matrix: (A aa) (atk fo Awl Ap AP App =] or -e Re Ke Ay Ad AS w -% Because of the transformation properties of the defining representation and its Conjugate: Ww aU, vi aw my, (4.154) the adjoint representation transforms as. Ay —> Ay = ULUSAE = (Ua) Aj (Uy). (4.155) 6, in terms of matrix multiplication, it has the simple form of Am Am O'AG. (4.156) Let us recall that here the U matrix is the defining representation of the SU(3) group.100 Group theory and the quark motel 4.14 SU(3) Invariant octet baryon-meson couplings ‘The baryon octet cun be represented by u 3 x 3 matrix ° ith B= ” 8 =F =n 7 Ve P and the pseudoscalar meson octet by Ringe reg (a) Construct the SU(3) invariant BBM couplings. (b) Express all the above meson-baryon couplings in terms of two SU(3) sym- metric couplings, This implies nunicrous coupling relations. But most of them actually follow from SU(2) isospin symmetry, Thus it is useful to express these SU(2) invariant couplings directly in terms of the two SUG) couplings. Solution to Problem 4.14 (a) In terms of these baryon and meson matrices, itis clear that there are only two invariant B BM couplings as there are only two independent traces of multiplying these matrices together: Liu = Vilgu tr BBM) + gor BMB)) = Vile (i! ato) + a2 (forty) ]. (4.15 A common way to express these couplings is to write me Stt, aoe, (4.158) so that Lao = rR RMD + DBM. BY. (4.159) v2 i Namely, the F coupling tx proportional to the commutator, and 2) to the anti- commutator. We now work out these couplings in terms of individual baryon and4.14 SUf3) variant octet baryon-meson couplings 101 meson fields: (D The m* couplings Le = VIM} (eB 8} + e281) = Vin (fF (3 %)* (- oe 5) eee | al(heF)e (G4) ]| = Vin (if & (rx- Dr) +8 oe (Fa-Ar-) + BS a (4.160) SP spp Or Lys = Vin! {aC (FFr*-r-) Be + nt. 4 DHE D- orf ‘ | (4.161) (h The x® coupling Lyo = V2 [mt (Bi 8! + gaditi) + M3 (wif n? +mi}n)] ="'(ul (Re 5)(Re q)tbe eee aA 5+%)-F | —=+ + Bet + bp ea) od) - [ta -a) (Ex —FFr) + ace OA + AE) +a (ES"- Bat) + atip - an). (4.162)Group theory and the quark mod | 4.14 Or Lye = afr (Er - Fert) + a (Ba +2’) + DoF (Fs - - Bis") + (nin). (4.163) (il) The K+ couplings Lx- = VIM} (eri B} + gril} BS) = VK Le P(5+4 * (4.164) ox = vane [PRE (E _2t* is-- 26 “a6 PA}: D+3F— D-3F (4.165) (lv) The K° couplings Lo = Vim} (a iB} + «:i}B}) = V2K°. {a [pe +8 (-5+ — Bw +m [Fe + (-5+ 5)° - van'[a( = + pe) +n/ = 281 + 82 a0 4 B1T 2e2 J peaks B29 nal. + Ree + A) a] (4 166)4.14 SU(3) imariant octet baryon-mesan couplings 103 Or Le = vans[Pe* (4.167) (¥) The n® coupling Lot = V3[M} («ud Bf + aalit a) + M2 (ff 82 + gol?) +3 («ff 8} + mdm) ] . ale (ia; + bf} - 2858) stan (Bi! Bt + BP BL — 28} a) ] + --- tin - 28-5" ~ 782 - fia] tee cone + (an — 2en) (S°E™ + BPE") - (2g - eG + an] [tn +m) (Fe +r" cg) + (Rit Raa v3 3F-D 2 ‘The other couplings are related by hermitian conjugation: Leake. Lhe me Le Lhe Lee. (4.169) [ (S=E> + De’ + BFE) - pan + (Fe + et). PHF op sin). (4.168)104 Group theory and the quark model 4.14 (b) The baryon octet contains the following isospin multiplets: singlet: A doublets: NV (*). (4.170) re triplet: £ = mee (ar- -r" while the meson octet contains singlet: 9° KY oe (RO doublets: K = (je): Kem (r-) (4.171) a Ane tripkt; l= Ge = ) From this it is straightforward to construct all the SU(2) invariant couplings, cf. Problem 4,8(b): (1) Three singlets nid (4.172) (i) One singlet @ two doublets NN, 985, KAS, KAN (4.173) (II) One singlet © two triplets nut Arf (4.174) (iv) One triplet @ to doublets Whn, Bhs, WExK, BEKe (4.175) (v) Three triplets rE En (4.176) By working out the components stich ax NEN = (ip - finn + VU pnnt + Apr) BEN wnt (BFE? —Dor-) - 2° (E"- Et) +0 (Erx- - Err) (4.17)AIS Isospin wave functions of two pions tos ‘and comparing them with the couplings shown above, we can easily express the ‘SU(2) invariant couplings in terms of the SU(a) 2 and F couplings. D-F RON) = oof, a(rEE) mF. (4.178) and similarly —RMAA) = gQVEE) = g(AL) D+3F 23 D-3F ROINN) = g(K AE) = REE) = (KAN) = 2V3 D+F 2 g(TER) = —g(K NE) = = (4.179) Remark. Just ax we have xeen in Problem 8 that, in terms of the ‘Cartesian com- ponents =(m 12 7), the coupling WTIN can be written as W rN + 7, the three triplet coupling ¢r £°£11 can be written as iE x E+ m, where we have used the identity CTT) F tune (4.180) Isospin wave functions of two pions Two pions can have total isospin / =: 2, 1, 0. Use the relations LILA) SIC +d be DIF AS 1) (4.181) HIB) SI = Wd + ht DEF DD (4.182) {0 construct the total isospin wave functions of two pions. Solution to Problem 4.15 (a) The = 2 stutes Starting from 12.2) = |xjty) (4.183) ‘we cun use f-[2, 2) = 212, 1) and #j1. 1) = V2}1.0) or ft!) = V2)r) wo get I BM Se (rntnd) + [nfs (4.184)106 Greup th oryand th quark mode 4.15 und use 1-2, 1) = 6}2. 0) and 1_|°) = JZ|n-) to get 12.0) = [se ez} + 2 [afr9) + [ae d)) (4.185) ' a Similarly, 1 7 = =I) = 5 (lntas) + [nr ag)) 12,-2) = |apaz). (4.186) (b) The / = | stutes #9} which is orthogonal tad) + b lint) (4,187) with lal? + [bI? = 1, the orthogonality condition becomes 1//2(a +b) = 0, The solution of ¢ = /VZ.can be chosen: 110) a (4.188) ‘This choive (as opposed to a = —b particular con- vention for the Clebsch-Gordon coefficients. It is easy t0 see that using the isospin lowering operator J_ we cun get the other # = | states: (4.189) (c) The / = 0 state ‘The [0, 0) state nust be orthogonal to both |2, 0) and |1. 0). This fixes it to be 1 — 10.0) = a (let az)— |xfx2) + [xe m})) (4.190) Remark, We note that the J = 2 and J = 0 states are symmetric under the interchange of particles | «+ 2, while the / = 1 states are antisymmetric, (This is why the combination |2r}72) is absent in the |1, 0) state.) In general, for two particles with the same isospin, the largest total isospin states are aymmetric under the interchange of purticles |<» 2. Then the next isospin states are untisynimettic, followed by synimetric stutes, etc. For exumple, in w system with two J = 3/2 particles, the state with 1 = 3, | is symmetric, while the one with 1 = 2,0 is antisymmetric.4.16. dsospins in non-leptonic weak processes 107 Isospins in non-leptonic weak processes ‘The low-energy AS = 1 non-leptonic weak Hamitonian is given by He = Fay = yold Iliya(1 — yada) + hee. (4.191) ‘The fist term (dad) in an isospin state ||, 1) and the second tern (#4) a |}, 4). From isospin addition of 0.0) ibd FEL d+ VFR (4.192) ‘we see that this weak Hamiltonian can be decomposed into two pieces with definite isospins: Hoe = Haya + Hyp (4.193) (a) Use this isospin decomposition of ty and Clebsch-Gordon coeflicients to evaluate the decay amplitudes for Kt-entn, Koontr, (4,194) in terms of two reduced matrix elements. (b) Repeat the same calculation for the decays A —* pr and A + n°, Solution to Problem 4.16 (a) Here we need to evaluate the mutrix elements of (rmlHelK) = Grmltaplk) + (er|Ayalk}. Since (K*, K°) is an isospin doublet, |K +) = |}. 44) and 1X9) = |}, —3), we have Maal) = [fy df) = 1h Hyak *) = |B 21h 3) = FI. — gL, (4.195) Hulk) = |b $)|$,- 3) = Fy, 0 + 10.0) Mapalk) = |$ 4) 1d 4) = Ap li2. 0) + 11, 0). (4.196) From the result obtained in Problem 4.15, 2.1) = Jy (lnfr9) + [nPmt)) U1) = Jp (ln n9) — [Paz (4.197) ‘we see that the final state of the decay K+ — +" must be either of the isoxpin states [2, 1) of [H, 1), But from angular momentum conservation, +7" must be in a relative orbital angular momentum L = 0 state, which is symmetric under108 Group theory and the quark model 4,16 the interchimge of m* «+ 1°, ‘Thus we must use the synimetrie combination as in 12, +1) and only the £ = 3/2 of the Hamiltonian can contribute: (a0 WH Kt) = Gr CT mm Halk *) =m (2, UW Aalfe $f). (4.198) The Wigner-Fekart theorem can then be used to relate it to a recluced matrix element Aya (2. tft] $$) = (2019 bf Aya = Bye (4.199) Namely, the decay amplitude is evaluated in terms of the reduced mutrix element as TUK ate) = PA. (4.200) As forthe decays K° -» +> and K® —» 77°, we have from Problem 4. 12.0) = Sp (pring) + 2 |mprg) + [ey ')) 1.0) = 55 (fitz) - [re af)) 10.0) = 4 (lafaz) - [yinf} + [ay ny) (4.201) which shows that the / = | state isantisymmetricin7* «> 97 is forbidden by angular momentum conservation and Bose statistics. We then have sy letnz) + bre ng) = f912.0) + fFi0.09 |apr$} = vino - Jie. 0}. (4.202) Thus (rt (Melk) = Jy ((2, 01 + V2(0, 01) Hy KD = Jy ((2. 0174521K) + V210. 0174y21K")) = ll. 0]}. fd. -H) Aya. + v2(0.0]3.§: 4. -H) Ara] = oh (yen - Aun) = fAn- pA (4.203) Or TK + n't) = hAia - fA (4.204) and similarly, TK" + nn) = SAva — BAU (4.205)4.16 fsaspins in non-leptonic weak processes 109 Remark, Experimentally, it has been observed that PUR > tn) & TUK > te) (4.206) TUK mtn) Lg TR ny (4.207) 29) This can then be translated through eqns (22), nd (22) into the anplitude relation of Ain «Ain (4.208) ‘This is the celebrated AJ = } rule of non-leptonic weak decay. (b) For the A + p> and A + n® decays, we have the isospin structure of A being an isosinglet, and tory = VN -H)- JED inn) = JEL B+ EE -H. (4.209) ‘Thus tom-abal = R$] Malay — fF. Hl tala = ce - a (4.210) (on°1rtetd) = VE. Moalay + PE. -$1 ala = ViAn + Via. (4.211) where Aya = (J. —$| HyalA) and Ai = (4 imental data are in agreement with the A of =$| 241214). Again the exper- $ rule Aya « Ara PI TUM pr) © T(A > nn) vi. (4.212)Chiral symmetry Another derivation of Noether's current Consider a system of scalar fiekls {gj}. = 1.2 depending on ¢, and Idi: L = L(di, Onds). Under an infinit dependent (local) transformation on ¢,: P(x) —> Om dy + Bor =o) + (XSL) (5.1) where (x) is some infinitesimal space-time-ependent purumeter and f,(6) is a function of the scular ficlds {@,}. (a) Show that the coefficient of @, in 5£ ix just the Nocther’s current us displayed in CL-eqn (5.15): with a Lagrangian al space-time al Wed = eS (5.2) (b) Show that the coetfticient of ¢ in 5 is then the current divergence d j,,(x). Solution to Problem 5.1 (a) Consider the variation of the Lagrungian: ac Lm 86, 8,0 (5.3) ago + Taapien. (5.3) Because variation and differentiation commute, and because 8¢, = € fi: Spb.) = MBG) = efi + Op fi (5.4) we have ac ac bl me! — fi + ———a, sail: (5.5) [54+ aes] + [saan] Noether's eurrent (for 2¢ = 0) in CL-eqn (5.15) ix just the coefficient of the Aye term in eqn (5.5). (b) The divergence of Noether's current can be evaluated directly: i. 3 =o" a mo 6) aut) = 8 waar [> sae |i Handy? * oo Using the Euler-Lagrange equation of motion, 7 of (32) fe 8 no, ads OGD5.2 Lagrangian with second derivatives ut we have a" f(x) = = +—— "fi, (5.8) ia ai which is just the coefficient of the € tertn of eqn (5.5). agrangian with second derivatives Consider the case where a Lagrangian depends on g, and aj, us well asthe second derivatives: LE Lib. Aybrs 9,06). (59) (a) Derive the Euler-Lagrange equution of motion for this cuse, (b) Derive Noether’s current for a global transformation: Gil) > O = Oe + 8G, = et fl) (5.10) where ¢ is some infinitesimal space-time-independent parameter. (©) Show that the current derived in (b) is the same as the coefficient of 3,€ in 88 =8 f Lax for local transformation. Hin: The higher order derivative termx yay can be reduced to 2y€ upon integration-| Solution to Problem 5.2 (a) The variation of the action Sm f 60d. Ie BOD being ss= {[H = 5b + Heonietion + 70, FeEg te] déx, (5.11) where we have used the property that differentiation und variation commute, Sati) = BBG.) and (By 3: 1 = 9, 9,(8G,). (5.12) ‘we cun obtain the Euler-Lagrange equation of motion through integration-by-parts: ac ar ee aa, a, 13) TORT) + TG OD 19) (b) For the global transformation, 54) = €f;(@). the vi becomes on of the Lagrungiun Oke: oe [it +a, Tear 20,846) sai. (5.14)2 Chiral symmetry 5.2 After using the equation of motion, and combining terns with the same nuinber of derivatives: ac BL me €8, py: . “at SMa aon! ae 7 wai saa aad ae [aon a4 A] +, [-ageeaa 4]. (5.15) 1 Cis invariant under the global transformation, the conserved Nosther current can then be identified through 8 = €2,.j" = 0: aL ac Saao' + i@aen AO)” HAO Oe fi (8.16) Remark. After applying the equation of motion 6£ can always be written as the divergence of some 4-vector (which is the essence of Nocther's theorem) because the equation of motion follows from 5S = f8Ld*x = 0 which comes about because terms having the sume number of derivatives combine into a total diver- gence so that it vanishes upon integration-by-pants, (©) Here we consider the local transformation Bh mex) fild) und, (5G)) me (Dye) fi + CB, Si. (5.17) In the variation of the action, 5S = -/[# 5g, + iG, wa (66) + Sa aan oe auatton atx ae -/ Lie f i+ as renin wh) ma, sana WEL +O Sr vfs. (5.18) the very lust term cun be rewritten upon ier aion by-parts ac ~ f eaagy obs =- [iegaaagtlits (Ew Si + yd Sd'x (5.19) -lmen 50 that all ternis are either a toe or ye: = =/ {Lie * waaay i OS + Tae wacasse) waa] Jets. 2m - [sacar - mace ihe + amen From thia, we soc that the coefficient of Aye is precisely Noether's current,5.3. Conservation laws in non-relutivistic theory u3 Conservation laws in a non-relativistic theory Consider a non-relativistic system described by a Lagrangian £ = L(qi. x) with qu. 1 = 1,2....0m, being the generalized coordinates, Suppose /. is invariant under the infinitesimal transformation Gg =H t beng). (5.21) (a) Show that the quantity (Noether's charge) given by ar Q= aq (5.22) is conserved, dQ/dt = 0. (bp If the Lagrangiun is given as L = $mv? V(r), where v = dr/dt with (1.42.43) and rc? = x7 + a} +23, show that / is invariant under the infinitesimal rotations xy ey bet, (8.23) where €) = —€y, Explicitly construct the conserved charges, (c) For the case where / = dmv’, show that / is invariant under the spatial translations , r—orertea (5.24) where a is an arbitrary constant vector. Find the conserved charges. (4) Consider a system of two particles interacting with cach other through a poten- tial which depends only on the relative coordinates V(ry — 2). Show that the total momenta p—miv1 + mg¥2 are conserved, Solution to Problem 5.3 (a) The variution of the Lagrangian is given by Arie JER aL... Bh ety 4 agg, (825) aq Using &%y = (cl/dt)(8qu) and the equation of motion ae _ dat 84, dt 8g, =0, (5.26) we get dab, | abd bi mw fy 4 wag tt wae a4 (ae d aL =a ia) sick (Fenn) (5.27) ‘Thus the invariance of the Lagrangian 8Z = 0 implies the churge conxervanon dQ/de = 0, with the charge being Q = (91/8q/)4;4,-4 Chiral symm try 5.3 (b) We can see that tee ete Og (5.28) 2 a follows from €y = —€), because ar x ' Br me Soba = bm me Sane) = 0 (5.29) and By? = 25/85) me 2 eyj% (5.30) We can use the expression in equ (5.27) d (aL d (al. a1 sey a) = hug, (ex - (5.31) to find the conserved churge for this rotational synunetry, aL ak : : yy = oy — Sx me may xy — mig, me Dixy — PLR (5.32) a) Bey which are just the familiar angular momenta, (c) Because a is a constant, we have v’ = v and thus 52. the charge of this spatial translational symmetry, we note that, for an infinitesimally small a, one has 8x; = a, and thus a (ab a (ab) _d teat (tess) wat (SH) wad omia. 2 a (ee) "dt () qt ee yuently 5L = (leads to the conserved charges m3, which are just the lineur momenta, (d) The Lagrangian for this case is given by ' 2 Clearly, £ is invariant under the spatial translation of the form Le jmut + smi — Vir, m2). (5.34) noreante momenta (5.35) wl translations, we have 5x, = 6rz == a.und set (Hew + zn) Ora dt \ary a ; = a> (my + mara). (5.36) a ‘Thus the total momentum Pam +mv2 (5.37) is conserved,54. Sym tries of the linear oom rdet us Symmetries of the linear 7 -model ‘The Lagrangian for the linear o-model is given by L= hyn)? + ,0)7] + Ni" = 2 oy tailo tinr mn + Slate x!) F(ot+ YP (538) where N = (") is an isospin-} nucleon field, #=(771, 72, 773) an isospin one pion ticld. and o an isospin zero scalar field. It is convenient to use a 2 x 2 matrix to represent the spin 0 tields collectively: Esotiren, (5.39) (a) Show that the Lagrangian is invariant under the isospin transformations: N—NSUN, E> P'=ULU, (5.40) where U = exp({o.- r) isan arbitrary 2 x 2 unitary matrix with o = (a, 2.04) being u set of real constants, Find the corresponding conserved isospin vector currents Vi, i = 1.2.3. (b) Show that the Lagrangian invariant under the axial isospin transformations . Bex N—Nmexp (iin) EE aVEv, (sah where V = exp({ B+ 1) is an arbitrary 2 x 2 unitary matrix with B= (1. fa.) being a set of real constants, Find the corresponding conserved axial-vector cur rents Ai, (©) Calculate the charge commutators l@'.0' Ie. 10“. OM (5.42) where = f PxVcx) und QY f PAGO). (4) Calculate the commutators of particle fields with the vector charges: WN (QR [Pal (5.43) and with the axiul-vector churges: (ON 10% 10%a1. (5.44)6 Chiral symmetry 5.4 Solution to Problem 5.4 (a) [tis useful to define the left-handed and right-handed chiral nucleon fields: (l-y)N and Nee tl +n. (5.45) Thus N= Nr + Np. ys = —Ne. and ye = +N. Also, E&ta(o tire mo—it m)sott(r mt a(ot WI (5.46) where we have used the Pauli matrix identity (r-A)(t- B) = (A.B) +i tA x B), (5.47) ‘We can then write the Lagrangian as Lo fer (a, 50" 5") + Np iy"O.Ny + Neiy"Ne . 2 a +R (NENE + REIN) + SUr(EE!) — TLE. (5.48) For the isospin rotations, we have Np —> Ny =UN;, Ne—> Ng @UNeg, aml D— L'=Urut. (5.49) Thus, (LE) = (UE \UEIU') = (EE) (5.50) and, in the same way, tr(2,, £3" EP) = tr(9Z9"EZ"), Also, Npiy"8uN; = N.Utiy4a,U Ne = Nyy" Ne. 650) and, sitnilarly, forthe Ngi y8j, Np term. Furthermore, Ny E'Ny = N,U'ULUTUNg = Ny EN p. (5.52) ‘Thus C is invariant under the isospin rotation, To get the conserved current, we need to work out the infinitesimal transformations: or Also, EE a (14152) term (1-15*) = otir-etP[S4, rx]. (5.55)5.4. Symmetries of the linear o-model 17 From [1,, ty] = 2ieyete We can work out the last commutator to be [$e] Siena sie x mr. (5.56) Thus, Eso’ tite Sotiv w-itax x) (557) or wen ax (5.58) Namely, Som0, Se =—ax x, (5.59) ‘The conserved isospin vector current Vj, is simply the Noether current for this symmetry transformation: al aL ac = aan tae) = syn — 8 x(a xm), (5.60) Or Wes fy -a" ax, (5.61) (b) The axial transformation of the nucleon field N— N'=exp (4s) N (5.62) takes on a simple form when expressed in terms of its chiral components Np= VN, and = Ne = VNR (5.63) where V = exp (42). It is then easy to see that both Ny iy"a,N;, and Neiy" Na are invariants, Sitilurly, both Ir(LE") = ar(VIEVINVETV)) = ar(EE") (5.64) and #r(9,, Lo" L*) are also invariants. For the Yukawa couplings, NEN, = NL V(VIEV VN = NENG. (5.65) Hence £ ix invariant under the axiul ixospin rotations. To get the conserved current, we need to work out the intinitesimal transformations, B n— Ween (Sty) ve (1HAtn) a (5.66)ea Chiral symmetry 5.4 or the infinitesimally small change of the nucleon field under axial transformation being b5N = Fan (5.67) Jeorie-m(t if x, tom} tee. (5.68) rare (- Z 2 atirn-ip- a+ 7 From (rj, t)} = 25y we can work out the lust commutator to be [Ex rrbe een (5.69) Thus, Oo pite nm Sat (BP m)ti(t- x)-io(B- t) (5.70) or . omo+Ben, amx-of. (5.71) Namely, Se] 6 = bee of. (5.72) ‘The conserved isospin uxial-vector current A, is siinply the Nocther current for this synumetry transformation: al aL —p- AY = ——, ieee eo ee = iy ae o(B-e)-cde-B. (5.73) ac a+ aah Ate Ayan = (xa"o -09"R), (5.74) Remark. We can combine the transformations in (a) and (b) as follows: Ne —> Ng RNe (8.75) Ni — Ni = LM, (5.76) EY a/ER (8.77)$4. Symm tries of the linear o-model he where we have introduced the right-handed and left-hand transformations. R mexp( ) and i=en(47) (5.78) with y = 8 = o for the vector transformation, and -8 = 8 for the axial transformation, Under the infinitesimal transformation, R = (1 + 13): Se =O (5.79) and z ont Ee (5.80) or oo = 2 bya = 7 bya Z 3° (5.81) From these field variations, we can immediately work out the corresponding con- served current Ree Hay" SNe - ; ‘Similarly for the left-handed transformation, L = (1 + 48+ 1): ato = 30nn x n=oI! HX), (5.82) ber BuNe=0 und BN, IN (5.83) and rs ( +2) tiem sine (5.84) 2 or bx xx 3 joe, yee Ta Se, (5.85) leading to the conserved current a uy in z Lt = Ney" ZN; + Soto — 5" x atod"m), (5.86) ‘The vector and axual-vector currents are then given as VomR +L" and AM = RY AL, (3.87)120 Chirat symmetry 5.4 (c) Let us first consider the vector charges Oe= f PxVi(x). (5.88) Because of current conservation, 3*V;! = 0, the churges Q* must be time- independent, We cun then choose to work with equal-time commutators: 12! Oligayy = fere'y [wide = umd’r'n™, ww ~ e’n'r!] (5.89) which cun be evaluated, see Problem 4.6, ax five ll tle Ul now a. [w w3neo.wordnon] wieo[$. 2] yow'a-v = in (WEN) Bax» (5.90) and st € jn Or! (x) (x), Oxy" eee yy Est nn (15 On" yx) + 189m (x) (y)) BK - y) =i (Chet PH — Emer) BK yd = 10a a! = alta" + n'a! — arn )B x= y) Heinen (Or! eB (Ky). (5.91) After substituting the results of eqns (5.90) and (5.91) into (5.89): 10! Ohm = fee fos (wtgw —eundx'n”) mi Qt, (5.92) Ina similar manner, we can verify the other commutation relations of (QQ) =tewQ* and 10%, OY] mie gt. (5.93) Remark. If we define the right-handed and left-handed charges as O=O+O, =a ov. (5.94) the algebra becomes [ei et ten Oh 595) tet, — [24-24] and [et a4] =o. (5.96) Namely, each set of {4 } and { Q%} separtely form an SU(2) algebra, This is why it in referred to us the SU (2), x SU(2)e algebra.5.4 Symmetries of the tmear o-model 11 (d) Here we calcutate the commutator of the various ficlds with the isospin charge 1 Q'= f wx (moze . eas! eon) . ($97) For the isodoubtet nucleon field, iy 10. Ny] = fex [ven (5) NW). won| (5.98) we can use the identity [AB,C] = A(B,C) = (A. C)B (5.99) toget yer lO. Ny) = - (5) N*(y). (5.100) For the isotriptet pion field, 1Q'.2'(y) ne f sian oan. x00) ~et [bs mi Matty), (5.102) Maxx -y) (5.101) For the singlet afield, lol e(] <0. (5.103) By comparing these results with the eqns (5.54) and (5.58) obtained in Part (a), we note that these charge-ficld commutators just yield the variation of the field under the isospin transformation with some parameter ot: [oe Q.G(xd] = 15d(x), (5.104) where $(x) = N(x), 1!(x), or (x), Similarly, for the axial charge, we have [8-Q. (0) = i860): 1\ae lo. Nn = (5) WN), (5.105) 10%. 2'(y)] = 184 a(y), (5.106) 1O* oy w ix(y), (5.107)122 Chiral symmeiry $5 pontaneous symmetry breaking in the 0 -model In Problem 5.4 the effective potential for the scalar ficlds has the form of i: a Va—Seott yt Feat whe (5.108) For the case of 4? > 0, the mininwm of this potential is at [see CL-eqn (5.168)], ain @emav, ve(F) (5.109) In the text, we chose the vacuum configuration to be fry = (nr) = ()=0, (a) =v. (5.110) Now consider the alternative configuration of (7) = 0, (rr (a!) = ry 6.11) (a) Show that the charges which do not annihilate the vacuum are (", Q?, and Q™, and the Goldstone bosons are m', 77, and ¢ fields. (b) Show that the remaining charges, Q*', Q*?, and Q?. form an SU(2) algebra. (c) Show that the fermion bilinear Ly, = guNiyst'W generated by (n°) = vcan be transformed into the standard fermion mass term of Lay = ty N'N’ by some chiral rotation, Find this transformation, Solution to Problein 5.5 (a) Given that (7) % 0, we seck charge-field commutators eqns (5.100)-(5.107) which are proportional to the 7? field: (Q'. 27) = -107.2'}= 10". 0) =i, (5.112) We see that the charges Q! the above equation would imply that (3) boson fields. {b) From the charge commutators calculated in Problem 4 we have lo". 71=10". -10. O"l=i07. 107.0" This means that the charges ("', QS, und Q* form an SU(2) algebra. (c) ‘To find the chiral rotation it is useful to decompose the fermion field into its chiral components, N= Ny + Np. In this way we have Q?, and Q* do not annihilate the vacuum [otherwise O), and 9}, 2, and a are Goldstone Qo". (5.113) Lum guNiyseN = gu[Nrit?Nx— Nal Ne]. (5.114) Consider the chiral transformation Np SEN}, Ne = RNR (5.115)5.6 PCAC In the o-model 123 where /, and R are unitary transformations. To turn Lan = gu [Np Ltit® RNg — NR LN; | (5.116) into the standard fermion mass term of — (A, Np + Ng; ), we require LiPRs-l, RiPL=t, (5.417) which is uctually one condition as (Ltiz3R)t = — RM solution 10 this condition, hence the required chiral rotation, is =1. One simple rt ow (-im). (5.118) where we have used the identity of 6 exp (:05;) = on§ +i (5.119) Remark. It spontaneous symmetry breaking, the choice of the vacuum expecta- tion value (VEV) direction is a matter of convention. All different choices yield- ing the same symmetry-breaking pattern are physically equivalent. In the example under discussion, both choices of (rr) 34 0 and (a) # O give the same symme- try breaking, SU(2) x SU(2)—» SU(2) or equivalently $O(4) + SO(3), and have exactly the same physical content. CAC in the o-model Suppose we introduce a synimetry-breaking term into the a-model Lagrangian Lsn = —co(x) (5.120) where ¢ is a constant, (a) Find the new mininwum for the effective potential, we a Vm Sot + a) + F007 + IY + €0. (5.121) (b) Show that in this case, pions are no longer massless und, in the tree level, their masses are proportional to the constant ¢. (©) Show that the axial-vector current A, derived in Problem 5.4 is no longer conserved. Calculate the divergence 84 A,, and show that it is proportional to the pion field x.124 Choral symmetry 5.46 Solution to Problem 5.6 (a) We have the minimization conditions x =u? +0? + elo + = 0 (5.122) and av oop = [w+ Mo? + m2) In! = (5.123) an Since it is not possible to huve (—? + 4(0? + x°)] = 0 for a non-vanishing c, we must have 27! = 0 and the afield satisfying the cubic equation: nia +o +e m0. (5.124) Remark. In this case the vacuum configuration is unique because the symmetry breaking term of eqn (5.120) hus singled out a direction. (b) To discover the physical content of the model, we shift the fields. wer o'ma-y, (5.125) where v is the solution to the cubic equation —17v + Av* + ¢ = 0. ‘The terms in the effeetive potential become att et sat + x2 4200! tu (5.126) (o? 4 PY me dvte? + 2v2(o? + #7) + non-quadrutic terms. ‘We then have the mass terms in the effective potential Van 2 (02 +0) +2 faut? + 020074 x)] = (aH) aay (2,2) go - (Fe Mats (But) 6.127 where to reach the last fine we have used the cube equution for v. Thus my = aut — (5.128) (5.129) If we pick ¢ < 0, then both m2 and m? are positive. That all three components. of x have cqual mass means thut the explicit breuking Ls = —co(x) still leaves isospin SUQ2) symmetry unbroken.56 PCAC in the a-model 125 (e) The divergence of the axial current is related to the variation of the Lagrangian as [see CL-eqn (5.14) or Problem 5.1] BBA, mds. (5.130) Ihe right-hund side would vanish were it not for the presence of the symmetry- breaking term £sp = —car(x), Uhus b5L =dsLsp = Fete =—cB RK, (5.131) where we have used a result obtained in Problem $.4, eqn (5.72). In this way we find OA, = cK. (5.132) Remark 1. The constant ¢ can be reluted to the pion mass and the pion decay constant, my and fy. For the x > j1v, deeay, the amplitude is proportional to the axial current matrix element, which defines f,, by (OAS (OR (p)) = 18" Fy Pus (5.133) Thus the matrix clement of the divergence is given by (O]d"AL(O)|2*(p)) = 8" fem? = —e(Olr'(0)]x*(p)). (5.134) Or = fami, (5.135) In this way, the divergence has the PCAC form BAS = famin (5.136) ‘The specific value of the pion decay constunt is fixed as follows, The amplitude forthe r+ — jy, decay can be written as T= Feeo14; ont (pyay"(1 = ysdv (5.137) with (014; OVI (p)) = iV2 fe Pre (5.138) With this definition, one finds from the decay rate that [see Problem 11.3(c) for the calculation] 3 Sa amy = 92MEV, (5.139) v2 Remark 2. Compuringeqnx (5.129) and (5 |35) we sec that the VEV of the a-field is simply the pion decay constant: (5.140)