F34TPP Theoretical Particle Physics notes by Paul San Contents

1 Lecture one 1.1 Natural units. . . . . . . . . . 1.2 converting back to SI units . . 1.3 relativistic notation . . . . . . 1.3.1 raising/lowering indices 1.4 examples . . . . . . . . . . . . 1.4.1 energy-mass relation . 1.4.2 quantum operators . . 1.4.3 Klein-Gordon equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 5 6 7 7 7 7 8 8 9 9 11 12 13 14 15 16 17 17 18 19 21 21 21 22 23 23 24 24 25 26

2 Lecture two - Dirac equation 2.1 Dirac equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 solving the Dirac equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Lecture three - spin, chirality 3.1 rotation generators . . . . . 3.2 rotating a fermion . . . . . . 3.3 chirality and helicity . . . . 3.4 Weyl fermions . . . . . . . . and helicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

4 Lecture four - gauge symmetry and electromagnetism 4.1 electromagnetism . . . . . . . . . . . . . . . . . . . . . . 4.2 gauge invariance of the Dirac equation . . . . . . . . . . 4.3 covariant derivatives . . . . . . . . . . . . . . . . . . . . 4.4 the standard model and gauge symmetries . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

5 Lecture ve - Dirac equation and magnetic elds 5.1 magnetic moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 magnetic moment of electron - Zeeman interaction . . . . . . . . . . . . . . . . . 6 Lecture six - Feynman rules 6.1 propogator theory . . . . . . . . . . . . 6.1.1 Schrdinger propogator . . . . . o 6.1.2 Klein-Gordon propogator . . . . 6.2 Feynman rules - including interactions 6.3 diagramatic expansion . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

7 Lecture seven - action functionals and Feynman rules 7.1 action and equations of motion . . . . . . . . . . . . . . . . . . 7.1.1 Klein-Gordon action . . . . . . . . . . . . . . . . . . . . 7.1.2 Dirac action . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 the structure of an action . . . . . . . . . . . . . . . . . . . . . 7.2.1 example, real KG action I . . . . . . . . . . . . . . . . . 7.2.2 example, real KG action II . . . . . . . . . . . . . . . . . 7.2.3 example, complex KG action . . . . . . . . . . . . . . . . 7.3 example, Dirac particle coupled to photons . . . . . . . . . . . . 7.4 complex KG equation coupled to photons . . . . . . . . . . . . . 7.5 Feynman rules for the electroweak sector of the standard model 7.6 Fermis golden rule . . . . . . . . . . . . . . . . . . . . . . . . . 7.6.1 ee scattering . . . . . . . . . . . . . . . . . . . . . . . . . 7.6.2 decay . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Lecture eight - group theory 8.1 denition of a group . . . . . . . . . . 8.2 direct product of groups . . . . . . . . 8.3 Abelian groups . . . . . . . . . . . . . 8.4 Lie groups (pronounced Lee groups) 8.4.1 general linear group GL(n,R) . 8.5 special linear group SL(n,R) . . . . . . 8.6 orthogonal group O(n,R) . . . . . . . . 8.7 special orthogonal group SO(n,R) . . . 8.8 O(p,q,R) . . . . . . . . . . . . . . . . . 8.9 symplectic group Sp(2n,R) . . . . . . .

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

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

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

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

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

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

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

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

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

26 26 27 28 28 29 30 30 31 31 31 33 33 34 34 35 36 36 36 36 37 37 37 38 38 38 38 39 39 39 40 40 40 41 42 42 45 46

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

9 Lecture nine - non-Abelian gauge theory 9.1 unitary groups . . . . . . . . . . . . . . . 9.2 unitary group U(n) . . . . . . . . . . . . 9.3 special unitary groups . . . . . . . . . . 9.4 unitary groups and the standard model . 9.5 generators and Lie algebras . . . . . . . 9.5.1 o(n), the Lie algebra of O(n) . . . 9.5.2 su(n), the Lie algebra of SU(n) . 9.6 non-Abelian gauge theory . . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

10 Lecture ten - some representation theory 10.1 SU(2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 normalizing the states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 general SU(2) represenation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11 Lecture eleven - isospin 11.1 neutrons, protons and isospin . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 pions and isospin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 using isospin in scattering calculations . . . . . . . . . . . . . . . . . . . . . . . 12 Lecture twelve - the quark model 12.1 more multiplets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Lecture thirteen - charge conjugation 13.1 Charge conjugation . . . . . . . . . . 13.1.1 photons . . . . . . . . . . . . 13.1.2 fermion-antifermion pairs . . . 13.1.3 C violation . . . . . . . . . . 13.2 parity . . . . . . . . . . . . . . . . . 13.2.1 electromagnetism . . . . . . . 13.2.2 fermions . . . . . . . . . . . . 13.3 linearity of parity . . . . . . . . . . . 13.4 electron dipole moment (edm) . . . . and parity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

46 47 48 49 50 50 54 54 55 55 56 56 56 57 58 58 59 59 60 61 62 62 62 63 65 65 65 66 67 68 69 71 71 71 72 73 73 75 75

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

14 Lecture fourteen - parity violation and time reversal 14.1 parity violation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.2 time reversal symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Lecture fteen - CP symmetry 15.1 CP symmetry . . . . . . . . . 15.2 neutral kaons . . . . . . . . . 15.3 kaons and C, P and CP . . . 15.4 Klong and Kshort decays . . . . 15.5 matter vs anti-matter . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

16 Lecture sixteen - quark mixing 16.1 three generations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2 quark mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Lecture seventeen - neutrino oscillations 17.1 Solar neutrinos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2 neutrino oscillations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.3 atmospheric neutrinos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Lecture eighteen - Higgs mechanism and 18.1 symmetry restoration/breaking . . . . . 18.2 superconductivity . . . . . . . . . . . . . 18.3 massive elds . . . . . . . . . . . . . . . 18.4 breaking the symmetry . . . . . . . . . . 18.5 non-Abelian Higgs mechanism . . . . . . 18.6 Yukawa coupling and fermion masses . . 3 symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . breaking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

19 Lecture nineteen - the real standard model 19.1 electroweak sector, part I . . . . . . . . . . . 19.2 chiral theories . . . . . . . . . . . . . . . . . 19.3 electroweak sector, part II . . . . . . . . . . 19.4 electroweak sector, part III . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

75 76 76 77 77

List of Figures
1 2 3 4 The predicted ux of solar neutrinos, Bahcall et. al. ApJ, 621, L85 (2005). . . . The typical shower of particles produced by a cosmic ray. . . . . . . . . . . . . . This shows a superconductor hovering above a magnet. . . . . . . . . . . . . . . A typical wine bottle, where the dregs of the wine tend to form at the bottom, around the circular minimum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 71 72 74

List of Tables
1 Table showing quark charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

F34TPP Theoretical Particle Physics 1 Lecture one

Aim: To introduce the notion of natural units, dimensional analysis, and revise some relativistic notation. Learning outcomes: At the end of this lecture you should be able to convert quantities between dierent systems of units know how to use Einstein summation convention

1.1

Natural units.

The rst thing to note is that the numerical value of a dimensionful quantity does not have any intrinsic value. For example, if you were told that the speed of light was 7.91014 , that would be meaningless; the reason for this is that the numerical value depends on the system of measuring rods and clocks you use. If you use metres and seconds you would get c = 3 108 ms1 , if you use miles and hours you get c = 6.7 108 miles hour1 , and if you use cubits and fortnights you get c = 7.9 1014 cubits f ortnight1 . Similarly, if we use kilograms, pounds or grains to measure measure mass we nd that = 1034 m2 kg s1 , 3 1037 miles2 lb hour1 , 9 1024 cubits2 grain f ortnight1 . So, by picking our units appropriately, we may essentially get any value we wish for dimensionful quantities. As we are interested here in relativistic quantum eld theory, it makes sense to choose units where c = 1, = 1, (1.1) (1.2)

these are called natural units. For example we may use the year as our measure of time, and the light-year as our measure of distance, then light travels one light-year per year, i.e. c = 1 in these units. In eect, we should think of c as the quantity that allows us to relate distances to times, and is the quantity that relates masses to times and distances. Using these natural units we nd that the basic relations E 2 = m2 c2 + p2 c2 , E = 2 p = become E 2 = m2 + p2 , E = , 2 p = . 5 (1.6) (1.7) (1.8) (1.3) (1.4) (1.5)

Now, we still need to pick our basic unit, and in particle physics the convention is to use electron-Volts, eV, to measure energy. Then, from (1.6), we have that mass is measured in eV , as is momentum. Now (1.8) shows that distances are measured in (eV )1 , and (1.7) gives that time is measured in (eV )1 .

1.2

converting back to SI units

At some point an experimenter will want to know how big their machine should be, and rulers dont come with eV notches on them, so we need to able to convert from our eV numbers back to SI units. The best way to see how this is done is to use an example, so consider the following. The scattering cross section, , of a target particle, and the scattering rate, , are related by = nv (1.9)

where n is the number density of particles that are being sent toward the target, and v their speed, so is the number of scattering events we expect per unit time, the dimensions are as follows [] = [L2 ], [] = L3 L2 LT 1 = T 1 (1.10) (1.11)

So, suppose we are told that the cross section associated with Compton-scattering (ei e ), in natural units, is = 8 2 2 . 3m2 e (1.12)

This clearly cannot be in SI units, because if it were, the left-hand-side would be m2 , while the right-hand-side is in kg 2 . Note that the ne structure constant 1/137 is dimensionless. So, to make sense of this expression we need to sprinkle some factors of and c around, so write 8 2 2 = 3m2 e
a b

c.

(1.13)

where a and b are unknown constants. Note that this gives the same expression as (1.12) in natural units where = 1, c = 1. Now we do some dimensional analysis [L2 ] = [M L2 T 1 ]a [LT 1 ]b [M 2 ] (1.14)

For this to be consistent, we see that a = 2, b = 2, so = 8 2 2 2 . 3m2 c2 e (1.15)

1.3

relativistic notation

First of all, recall that the object variously called the line element/proper distance/proper time/invariant interval... is given by ds2 = dt2 + dx2 + dy 2 + dz 2 . (1.16)

Rather than writing this out each time we introduce a matrix quantity called the metric, with components , where the rst index labels the rows, and the second index the columns of the matrix. In (t, x, y, z) := (x0 , x1 , x2 , x3 ) co-ordinates the matrix is diagonal, with entries = diag(1, 1, 1, 1). Given this, the line element is ds2 =

(1.17)

dx dx .

(1.18)

where we have introduced the spacetime four-vector with components dx = (dt, dx, dy, dz). The summation convention simply says that if an index appears twice in a single expression, then that index must be summed over. So, for example, the line element becomes ds2 = dx dx . There is no need to explicitly write the know they must be summed over. 1.3.1 raising/lowering indices

(1.19)

symbol, because both and appear twice, so we

Along with the metric we have the inverse metric , which is just the inverse matrix of the metric. So, in (t, x, y, z) co-ordinates we have = diag(1, 1, 1, 1). This allows us to raise and lower indices of spacetime vectors, for example V = V V = V . (1.21) (1.22) (1.20)

This can be useful as it often further simplies certain expressions, meaning that we dont even need to write , on top of dropping the summation symbols.

1.4
1.4.1

examples
energy-mass relation

recall that the energy-momentum four-vector is P = (E, p) 7 (1.23)

i.e. the zero-component of four-momentum is just the eneryg, P 0 = E. Using P = P we also have P = (E, p) Now we take the energy-momentum relation E 2 = p2 + m2 and rewrite this as E 2 + p2 + m2 = 0 (P 0 )2 + (P 1 )2 + (P 2 )2 + (P 3 )2 + m2 00 P 0 P 0 + 11 P 1 P 1 + 22 P 2 P 2 + 33 P 3 P 3 + m2 P P + m2 P P + m 2 = = = = 0 0 0 0 (1.26) (1.27) (1.28) (1.29) (1.30) (1.25) (1.24)

As this nal expression contains only spacetime indices it is manifestly a relativistic equation. 1.4.2 quantum operators

From quantum mechanics we know that E = it P x = ix where we denote x = . x (1.33) (1.31) (1.32)

Using the fact that P0 = P 0 and P1 = P 1 we see P0 = i0 , P1 = i1 P2 = i2 P3 = i3 or, using our relativistic notation P = i 1.4.3 Klein-Gordon equation (1.35) (1.34)

We have already seen that P P + m 2 = 0 and P = i 8 (1.37) (1.36)

so we may combine them to give m2 = 0 This is the Klein-Gordon equation. If we unravel the index structure we nd m2 = 0 0 0 + 1 1 + 2 2 + 3 3 m2 = 0 + 2 m2 = 0 where a dot denotes time derivative = t (1.42) (1.39) (1.40) (1.41) (1.38)

Lecture two - Dirac equation

Aim: To introduce the Dirac equation. Learning outcomes: At the end of this lecture you should know how to derive the Dirac equation know what the Dirac algebra is be able to solve the Dirac equation

2.1

Dirac equation
2

The Schrdinger equation o it + 2m

(2.43)

treats time and space on an unequal footing, and so cannot be relativistic. The Klein-Gordon equation on the other hand is manifestly relativistic, however, it is second-order in time derivatives, unlike the rst-order time derivatives of the Schrodinger equation. Dirac wanted to construct a relativistic wave equation that was rst-order in time derivatives, as well as being relativistic, so he wrote down the following 0 0 + 1 1 + 2 2 + 3 3 + m = 0 (2.44)

which is the most general linear, homogeneous, rst-order equation, where the are as yet unknown. This equation is written in relativistic form as + m = 0 In order to nd out what the are we use = iP and nd (i P + m) = 0 9 (2.46) (2.45)

 Given that this vanishes, we can also act on it with (i P m), which when acting on zero gives zero, so (i P m)(i P + m) = 0 ( P P m2 ) = 0 1 [ + ]P P + m2 = 0 2 (2.47) (2.48) (2.49)

Now, we denote the anti-commutator with curly braces { , } := + so we nd 1 { , }P P + m2 = 0 2 which we should compare to the relativistic energy-momentum equation P P + m2 = 0 leading to the conclusion that { , } = 2 . So, how do we use this to nd out what the are? Well, lets start with { 0 , 0 } = 2 00 ( 0 )2 = 1 0 = i Now lets try { 1 , 1 } = 2 11 ( 1 )2 = 1 1 = 1 And now we go for { 0 , 1 } = 2 01 0 1 = 0 i = 0 (2.56) (2.55) (2.54) (2.53) (2.52) (2.51) (2.50)

and we run into a contradiction. The reason this has gone wrong is that we assumed the are numbers, they are not, they are matrices, and so we should really write { , } = 2 I. (2.57)

This is known as the Dirac algebra. In fact, the smallest set of matrices that can satisfy the Dirac algebra are 4 4 matrices, meaning that is actually a four-component object - four complex components - as opposed to the Schrdinger wave function which is a single complex o function. In fact, there are innitely many choices of matrices that satisfy (2.57). This is not a problem, and in fact is a good thing; we can think of each of these dierent choices as being required for the innite variety of observers, all travelling at dierent velocities. For example, 10

suppose that a set solves (2.57), then so will = , and so will = T , and so will = . In fact, if we take a general invertible matrix R, we have that = R R1 will also satisfy the Dirac algebra, e.g. { , } = = = = = = + R R1 R R1 + R R1 R R1 R R1 + R R1 R( + )R1 R(2 )R1 2 (2.58) (2.59) (2.60) (2.61) (2.62) (2.63)

Although each of these dierent representations will lead to the same physical results, there are certain choices that can make calculations easier, two common choices are the: Dirac representation 0 = iI 0 0 iI , = 0 i i 0 , 5 = 0 I I 0 (2.64)

where the are the Pauli matrices. And the Weyl representation 0 = 0 iI iI 0 , = 0 i i 0 , 5 = I 0 0 I (2.65)

Now, you will have noticed that I sneaked in a new matrix, 5 . It turns out that 5 plays a crucial role in much of the physics of the standard model, and it is dened as follows, 5 = i 0 1 2 3 . And the remarkable property that it satises, making it such a useful object, is 5 = 5 . We shall have more to say about this later. (2.67) (2.66)

2.2

solving the Dirac equation

The Dirac equation is linear, so if we are to look for wave soluitons it is natural to look at the Fourier modes; that means we look for solutions of the form = eip x

UA UB

(2.68)

where UA,B are two-component column vectors. Given this ansatz we note that it = P0 = E 11 (2.69)

and if we use the Dirac representation we nd that EUA .P UB + mUA = 0 EUB + .P UA + mUB = 0 (2.70) (2.71)

In particular, if we focus on the case of a wave in the z-direction, P = (0, 0, P z ), we have (E + m)UA = P z 3 UB (E m)UB = P z 3 UA So that we nd (E 2 m2 )UB = (E + m)(E m)UB = P z 3 (E + m)UA = P z 3 P z 3 UB = (P z )2 ( 3 )2 UB = (P z )2 UB (E 2 m2 P 2 )UB = 0 (2.74) which is just the relativistic energy-momentum equation, as expected. As UB is a two-component object, we have two linearly independent solutions: solution one has z P /(E + m) Pz 0 1 1 = eiP.x (2.75) UB = UA = 1 0 E+m 0 0 solution two has 0 1 Pz E+m 0 1 0 P z /(E + m) = eiP.x 0 1 (2.72) (2.73)

UB =

UA =

(2.76)

In other words, for a particle travelling in the z direction there are two possible solutions, these are just the spin-up and spin-down posibilities, as we shall see.

Lecture three - spin, chirality and helicity

Aim: To see where angular momentum ts into the Dirac equation Learning outcomes: At the end of this lecture you should know what the angular momentum operators for a Dirac fermion are know what chirality and helicity are

12

3.1

rotation generators

By now you will have come across the phrase a fermion is spin-half a number of times over the course of your degree, here we shall see why. To do this we need to understand how to rotate objects, in particular fermions. The key is the angular momentum algebra, as it is the angular momentum operators that are resonsible for rotating objects, Lx , Ly , This is typically derived using Lz = xPy yPx Lz = ixy + iyx (3.78) = iL z (3.77)

relevant for orbital angular momentum, but in fact the L appearing in (3.77) can take a number of dierent forms. To see this, lets consider how we rotate a three-vector V , V =RV where R is a matrix satisfying RRT = I. For example, a is accomplished by cos sin Rz = sin cos 0 0 (3.79) rotation about the z-axis with angle 0 0 , 1

(3.80)

whereas a rotation about the x and y axes come from 1 0 0 cos 0 sin , 1 0 Rx = 0 cos sin , Ry = 0 0 sin cos sin 0 cos

(3.81)

Now we know how to rotate vectors (we just use the R matrices), what about rotating fermions? We cannot use R to rotate a four-component fermion eld , as R is a 3x3 matrix. In order to progress we need something more general, and the route to generality comes by again looking at the rotation of vectors, but from a dierent perspective. We write, for example, Rz = eiT where T3 0 i 0 = i 0 0 , 0 0 0 (3.83)
3

(3.82)

and the exponential of a matrix is dened through the taylor expansion of the exponential function, 1 1 eM := I + M + M 2 + M 3 + ... 2 3! 13 (3.84)

One makes similar denitions for Rx 0 1 0 T = 0

= eiT and Ry = eiT 0 0 0 0 i , T 2 = 0 i i 0

where 0 i 0 0 , 0 0

(3.85)

This may seem like an obscure thing to do, but the point is that general group elements, g, can be written as g = ei T , where the set of matrices T are known as the generators of the group. The magic thing is that, by explicit computation, T 1, T 2 = iT 3 , (3.86)

which is just the angular momentum algebra, (3.77). The point is that it is the angular momentum algebra generators that are resposnsible for rotations, be they orbital angular momentum, in which case one should use the generators given by (3.78), or vector rotation, in which case one should use (3.83,3.85). Our task now is to nd a set of 4x4 matrices that satisfy (3.77) so that we can rotate fermions.

3.2

rotating a fermion
S 1, S 2 = iS 3 , (3.87)

What we need to do is nd some set of matrices, S that obey the angular momentum algebra

these should be 4x4 matrices, in order to be able to act on the four-component fermion eld . As we already have a set of 4x4 matrices - the Dirac matrices - this is the natural place to start looking. In fact, we shall simply state that these generators are given by dening a set of matrices jk , where jk := S i := for example S3 = along with i S 2 = 31 2 i S 1 = 23 2 (3.91) (3.92) 1 2
312

i j k , , 4
ijk

(3.88) (3.89)

1 2

jk

12 +

1 2

321

i i 21 = 12 = ( 1 2 2 1 ) = 1 2 4 2

(3.90)

One may check, by explicit computation using te Dirac algebra, that (3.87) is satised. So, we now know how to rotate a fermion. For example, in the representations given previously we nd that the generator of a rotation about the z-axis is S3 = 1 2 14 3 0 0 3 , (3.93)

so, for example, our plane-wave solutions (2.75), (2.76) have 1 S 3 = + 2 1 S 3 = 2 (3.94) (3.95) (3.96) which, as expected, shows that has spin + 1 , while has spin 1 . 2 2 By analogy with vectors, a rotation of a fermion about the z-axis through angle is accomplished by = eiS ,
3

(3.97)

which means we may check explicitly the common phrase that a fermion picks up a minus sign when rotated through 2, by noting that eiS
3

= =

ei/2 0

0 e
i/2 3

(3.98) (3.99)

I cos(/2) i 3 sin(/2) 0 0 I cos(/2) i 3 sin(/2)

so, for = 2, = (3.100)

3.3

chirality and helicity

due to the presence of a 5 matrix we have a neat way of splitting up a Dirac fermion using the projectors PR = 1 1 (I + 5 ), PL = (I 5 ) 2 2 (3.101)

and we dene the left and right chiral components by R = PR , L = PL This can always be done, because = 1 1 1 1 + 5 + 5 2 2 2 2 = R + L (3.103) (3.104) (3.102)

we then note that the chiral components are eigenfunctions of the 5 matrix 5 R = +R 5 L = L 15 (3.105) (3.106)

At rst sight, there seems to be little reason to label these chiral components as left and right, why not white and black, chalk and cheese...? To understand this we need to introduce the notion of helicity. We now know that a fermion has two vector associated with it, its momentum, and its spin. Helicity tells us whether they are pointing in the same direction or not. If the momentum and spin are aligned (positive helicity) then the fermion is spinning in the direction given by the usual right-hand-rule, so we call it right-handed. We shall now see that this ties in exactly with the chirality of massless elds. The Dirac equation for a wave mode of a massless fermion may be manipulated as follows = 0 P = 0 0 P0 + .P = 0 0 P0 .P 0 = 0 5 P0 5 .P 0 5 = 0 P0 5 + .P i 1 2 3 = 0 P0 5 + iP x 2 3 + iP y 3 1 + iP z 1 2 P0 5 2P x S x 2P y S y 2P z S z E 5 2P .S P .S E = |P | |P | so we dene the helicity operator h := P .S |P | (3.108) = 0 = 0 = 0 1 5 = 2

(3.107)

which measure whether the momentum is aligned with the spin. So, we see that right and left massless chiral elds have 1 hR = + R (3.109) 2 1 hL = L (3.110) 2 which is why they are called right and left, they follow the right hand rule, or left hand rule. In the above derivation we made critical use of the masslessness of the fermion, this is because helicity is not Lorentz invariant for massive elds. For example, suppose we see a fermion with helicity + 1 , we can simply convert it to 1 by travelling faster than it in its direction of motion, 2 2 then P will change sign, but S will remain the same. We cannot overtake massless elds, and so the helicity of massless particles is a Lorentz invariant. However, the notion of chirality, as dened by 5 , is Lorentz invariant. Chirality and helicity only coincide for massless particles, nevertheless, we still call chiral fermions left and right, irrespective of whether they are massless.

3.4

Weyl fermions

We have just seen that mass plays an important role for chiral fermions, this can be seen in another context - Weyl fermions. Taking the Dirac equation and operating on it with PR , and 16

PL we nd PR [ + m] = 0 L + mR = 0 and PL [ + m] = 0 R + mL = 0 (3.113) (3.114) (3.111) (3.112)

If m = 0 then these are two independent equations, and we may consistently set either R = 0 and L = 0, or L = 0 and R = 0. However, if the fermion has non-zero mass we need both left and right components1 .

Lecture four - gauge symmetry and electromagnetism

Aim: To see what electromagnetism has to do with symmetry Learning outcomes: At the end of this lecture you should be able to turn a global symmetry into a local one know what role gauge elds play in local symmetries.

4.1

electromagnetism

In the standard model, the list of particles can be split up into matter particles and force carriers. Matter particles are the bits that stu is made out of, such as electrons, protons and quarks, whereas force carriers are what bind the matter particles together, such as photons and gluons. Here we shall see that the origin of the photon is a symmetry principle. First of all, what do we mean by a symmetry? The type of symmetry that we will be looking at is the analogue of a symmetry in Newtons second law, F = V. We can see quite clearly that this system of equations is unchanged if we take V V + V0 , V0 R (4.116) (4.115)

i.e. V0 is a real constant. This is a symmetry; we can shift the potential by any number, and none of the physics changes. A similar thing happens in electrical circuits, the current between two points is determined by the dierence in potential between those points, I=
1

V R

(4.117)

There is another type of mass term that we will not be considering, a Majarona mass, for which these conclusions are a bit premature.

17

so we can again shift the potential by a constant amount, everywhere on the circuit, and the physics remains the same. In electromagnetism we have that B = A, E = A. (4.118)

This system is again governed by derivatives, so we are free to shift both and A by constants without changing the physics; in fact, we can do much more than that. Rather than just shifting by a constant, we can shift by an amount determined by an arbitrary function, (t, x) , A A + and the physics remains the same, i.e. under these shifts E E, B B. (4.120) (4.119)

These shifts are known as gauge transformations. The fact that and A are undened up to this arbitrary function highlights that they are not physical quantities, only the electric and magnetic eld can be measured. In line with our relativistic notation we shall now introduce the four-vector potential, by combining and A into A = (, A), A = (, A) Then, in this unied language, we see that our gauge transformations become A A + . (4.122) (4.121)

This will make it easier when we come to the relativistic Dirac equation - which is now.

4.2

gauge invariance of the Dirac equation

The Dirac equation (4.123)

has what is called a U(1) symmetry - more of which later. Simply put, this means the equation for and are exactly the same if = ei , R (4.124)

so, is a real constant. This just means that we are allowed to perform a phase rotation by angle , so long as we do it everywhere in spacetime. This is what is termed a global symmetry - we do it everywhere. Now we ask the question, can we make this symmetry local, i.e. let (x)? The naive answer is no. For = ei(x) 18 (4.125)

we nd that the Dirac equation for implies + m = i , (4.126)

which is not the Dirac equation. Rather than just giving up, we notice that if, instead of starting with the Dirac equation, we start with ( iqA ) + m = 0 and A then we end up with ( iqA ) + m = 0. (4.130) = ei(x) , 1 = A q (4.128) (4.129) (4.127)

i.e. the (, A ) variables satisfy the same equation as the ( , A ) variables, moreover, the transformation of the four-vector A is exactly what is required by electromagnetism if we identify = /q. Let us recap. We now have something that looks a bit like the Dirac equation, but it generalizes the global U(1) symmetry of the basic Dirac equation to a local U(1), i.e. we can now perform a dierent U(1) transformation at each point in spacetime and the physics is unchanged. The penalty for achieving this is that we had to introduce a new quantity A , but thats just electromagnetism, so its more of a prize than a penalty! U(1) gauge symmetry implies the existence of a photon, A .

4.3

covariant derivatives

The best way to think about gauge invariance is through the notion of a covariant derivative. The reason that the original Dirac equation does not have a local U(1) symmetry is that when we take ei(x) , and transform dierently. What we really want is some derivative, D , that transforms as D ei D , (4.131)

because then the derivative piece of the enhanced Dirac equation ( D ) will transform in the same way as the mass term. This is just what we have constructed, D = iqA and we have = ei(x) , A = A + (D ) = ei(x) (D ). 19 (4.133) (4.134) (4.135) (4.132)

With this covariant derivative we are now in a position to construct new objects, the most important of which is the eld strength, F . This is dene by [D , D ] = iqF . (4.136)

This is a nice quantity because it is gauge invariant, and so could be of physical relevance. To see that it is gauge invariant we note that as D transforms covariantly, do does D D , that is one of the strengths of using covariant derivatives. This implies that [D , D ] also transforms covariantly (i.e. [D , D ] ei(x) [D , D ] ). But, the right-hand-side of (4.136) also picks up a ei(x) , meaning that F must be unchanged, F = F . To be more explicit note that = U , (D ) = U (D ), (D D ) = U (D D ) (4.137)

where U = ei(x) . So, the primed eld strength is just ([D , D ] ) U [D , D ] U (iqF ) U F Now, this is required to hold for all so U F = F U and now note that U U = 1 = U so multiply on the right by U . U F U = F (4.143) (4.142) = = = = iqF iqF U iqF U F U (4.138) (4.139) (4.140) (4.141)

then note that in this case the gauge theory is Abelian, i.e. U(1), so the U , U are just numbers and can move through the F to cancel each other F = F (4.144)

This proves the gauge invariance of the eld strength. Now we need to see what the eld strength actually is, to do this, just calculate it iqF = D D D D = ( iqA )( iqA ) ( ) = iq( A A ) so we nd F = A A (4.148) (4.145) (4.146) (4.147)

and now we are nearly there, just a few examples should convince you that F is nothing more than a re-packaging of the electric and magnetic elds. e.g. F01 = 0 A1 1 A0 = Ax + x = E x . e.g. F12 = 1 A2 2 A1 = x Ay y Ax = B z . It is a useful excercise to check the other components. 20

4.4

the standard model and gauge symmetries

In later lectures we shall come across more general symmetry groups than U(1); ones of particular importance are SU(2) and SU(3). The reason for their signicance is that the standard model is a gauge theory of U(1)SU(2)SU(3), with the U(1)SU(2) supplying the photon, the Z0 , and the W gauge bosons, and the SU(3) giving us the gluons.

Lecture ve - Dirac equation and magnetic elds

Aim: To see how electromagnetism ts in with the Dirac equation Learning outcomes: At the end of this lecture you should know what a magnetic moment of a particle is know how to derive the magnetic moment of a non-relativistic Dirac particle

5.1

magnetic moments

Recall from electromagnetism that the magnetic moment of a circular, planar loop of area A, with current I, is given by = IA. (5.149)

Now, for a total charge of q running around the loop, in period T we have v (5.150) I = q/T = q 2r so we nd q q mvr = L = (5.151) 2m 2m where L is the angular momentum of the current, and m its mass. Now, suppose we place such a magnetic moment in a magnetic eld, one nds that they prefer to align with the background eld. This is because aligned magnets have a lower energy. In fact, classical electromagnetism says that the energy of a magnetic moment, , within a magnetic eld B is given by the interaction Hamiltonian q Hint = .B = B.L (5.152) 2m This is the starting point for writing down the interaction energy of a particle of spin S within a magnetic eld. As we expect there to be dierences with the basic loop of wire we adopt a fudge-factor, g, to account for dierences, giving the interaction Hamiltonian as q H = g B.S (5.153) 2m where g for the electron is found, by experiment, to be gelectron 2.002 It is our task to derive the magnetic moment of the Dirac particle. 21 (5.154)

5.2

magnetic moment of electron - Zeeman interaction

Now that we have learned all about gauge invariance, we shall see one of its consequences, namely a prediction for the magnetic moment of Dirac particles. We start with the gauged Dirac equation ( iqA ) + m = 0 and we write the Dirac eld as = 1 2 (5.155)

Taking the Dirac representation of gamma matrices (2.64), the denition of the four-vector potential (4.121), and the relation between momentum and derivatives (1.35) we nd it 1 = (m + q)1 + (p qA).2 it 2 = (m + q)2 + (p qA).1 . (5.156) (5.157)

For a free wave, we have that the wave function varies as eiEt , and now we take the nonrelativistic approximation that E m, and use this in (5.156) to nd 1 1 (p qA).2 2m (5.159) m >> q (5.158)

showing that 1 is small compared to 2 , and also we nd that (5.157) now becomes it 2 = (m + q)2 + and we then use that [(p qA).]2 = (p qA)2 qB. which gives the nal equation (p qA)2 q 2 + (m + q)2 B.S2 = it 2 2m m (5.162) (5.161) 1 [(p qA).][(p qA).]2 2m (5.160)

which is just the Schrdinger equation for a particle in a magnetic eld, whose magnetic intero action is q Hzeeman = B.S (5.163) m giving a g-factor for the Dirac particle of gDirac = 2 22 (5.164)

As this is so close to the observed value for the electron, we identify the electron as a Dirac particle. The corrections to the value of 2 is explained using quantum eld theory - very precisely g2 2 = (1159.65218111 0.00000074) 106
experiment

(5.165) (5.166) (5.167)

g2 2

= (1159.65213 0.000003) 106

Lecture six - Feynman rules

Aim: To uncover the origin of a particuarly useful perturbative technique Learning outcomes: At the end of this lecture you should know what a propogator is be familiar with the route from Huygens principle to the propogator equation be able to solve the propogator equation when there is a small perturbation present be able to derive the propogtor for the Klein-Gordon equation

6.1

propogator theory

Based on an idea of Dirac, Feynman decided to formulate his own version of quantum mechanics, and he based it on Huygens principle for calculating the propogation of wavefronts. This should not come as a great surprise, given that the Schrdinger equation is a wave equation. The o question we want to ask is, what is the probability amplitude at some location xf , at time tf , given the amplitde at some earlier time t? To answer this, we introduce the notion of the propogator, G(tf , xf ; t, x), which tells us how an amplitude at (t, x) evolves into an amplitude at the late time (tf , xf ). (tf , xf ) =
x

iG(tf , xf ; t, x) x(t, x),

tf > t

(6.168)

What this equation does, is split space into blocks of size x, then the block at location x contributes x(t, x) to the amplitude at the initial time t. This disturbance then propogates to (tf , xf ) with the help of the propogator. We then have to add up all such contributions. The continuum version changes the summation to an integral, and we may incorporate the tf > t condition by including a theta-function (tf t)(tf , xf ) = dx iG(tf , xf ; t, x)(t, x), (6.169)

To see how this works in practise, lets take write the Schrdinger equation in the following o form, introducing the operator O O(tf , xf )(tf , xf ) = [itf H(tf , xf )](tf , xf ) = 0 23 (6.170)

 Now operate on both sides of (6.169) with O, using tf (tf t) = (tf t) to nd i(tf t)(tf , xf ) = which implies that O(tf , xf )G(tf , xf ; t, x) = (tf t)(xf x). (6.173) dx O(tf , xf )iG(tf , xf ; t, x) (t, x) (6.172) (6.171)

So, this just tells us that the propogator, G(tf , xf ; t, x), is the Green-function associated to the Schrdinger equation. What this means is that the propogator is, in some sense, the inverse of o the Schrdinger equation, i.e. OO1 = I. o 6.1.1 Schrdinger propogator o

The propogator for the free (no interaction terms in the Hamiltonian) Schrdinger equation o comes from solving [itf H0 (tf , xf )]G0 (tf , xf ; t, x) = (tf t)(xf x) (6.174)

As this is a linear equation we consider a solution using the Fourier transform by writing G0 (tf , xf ; t, x) = and then we use dxeix(ya) = 2(y a) to nd G0 (E, p) = 1 E p2 /2m (6.177) (6.176) dE dp G0 (E, p)eiE(tf t)+ip(xf x) 2 2 (6.175)

In fact, it is more common to call this, the Fourier transformed version, the propogator. 6.1.2 Klein-Gordon propogator

The Klein-Gordon equation is ( + m2 ) = 0 (6.178)

and to nd the KG propogator we simply replace with G0 (tf , xf ; t, x), and the zero on the right-hand-side by (tf t)(xf x). ( + m2 )G0 (tf , xf ; t, x) = (tf t)(xf x) 24 (6.179)

Again, the equation is linear so we look for the Fourier solution G0 (tf , xf ; t, x) = d4 p G0 (p)eip.(xf x) (2)4 (6.180)

and now we nd the propogator for the KG equation to be G0 (p) = 1 . p2 + m2 (6.181)

where p2 = E 2 + p2 is the spacetime-square of the momentum four-vector. Now we start to spot a pattern, the (momentum-space) propogator is just 1/(equation of motion), both for the Schrdinger equation, and the KG equation - this is rather generic, especially given that we o expect the propogator to be the inverse of the equation of motion, OO1 = I.

6.2

Feynman rules - including interactions

So far we have not discussed the propogators for interacting theories, only the free versions, now we shall add a small perturbation to the Schrdinger Hamiltonian to see how we may o accomodate interactions. The starting point is O(tf , xf ) = it H = it H0 VI = O0 (tf , xf ) VI O(tf , xf )G(tf , xf ; t, x) = (tf t)(xf x) These may be rewritten as O0 (tf , xf )G(tf , xf ; t, x) = (tf t)(xf x) + VI (tf , xf )G(tf , xf ; t, x) the solution of which is an integral equation G(tf , xf ; t, x) = G0 (tf , xf ; t, x) + dt1 dx1 G0 (tf , xf ; t1 , x1 )VI (t1 , x1 )G(t1 , x1 ; t, x) (6.185) (6.184) (6.182) (6.183)

This may not be apparent at rst sight, and is easier to conrm by working backwards. If we operate on (6.185) with O0 (tf , xf ), the left-hand-side simply recovers the left-hand-side of (6.184), while the rst term on the right-hand-side gives the rst term on the right-hand-side of (6.184). To understand the rest, note that O0 (tf , xf ) only sees tf and xf , its not interested in t, t1 , x, x1 , so when O0 (tf , xf ) hits the integral on the right-hand-side of (6.185) it just turns the G0 (tf , xf ; t1 , x1 ) into (tf t1 )(xf x1 ), allowing the integral to be performed. We now use an unusual trick, that usually results in 0=0 when tried for most equations, we substitute the equation (6.185) into itself, i.e. the replace the G(t1 , x1 ; t, x) inside the integral on the right-hand-side of (6.185) by the solution for G(t1 , x1 ; t, x). Dropping the details we have G = G0 + G = G0 + G = G0 + G0 VI G G0 VI G0 + G0 VI G0 + 25 G0 VI G G0 VI G0 VI G

Now we repeat what we have just done, by replacing the the G on the right-hand-side of this expression with the initial solution G = G0 + = G0 + G0 VI G0 + G0 VI G0 + G0 VI G0 VI G0 + G0 VI G0 VI G0 + G0 VI G G0 VI G0 VI G0 VI G

at which point we start to see whats going on. In principle we should do this an innite amount of times, but as we assume VI is small, the later terms will be negligible, meaning that we have an expression for the full propogater G in terms of the much simpler free-propogater G0 .

6.3

diagramatic expansion

As is common with such perturbative expansions, it is useful to use diagrams as mneumonics for the integrals, and objects appearing in the integrals. The diagramatic expansion for the above solution to the integral propogator equation is

f ull propogator, G =

where the line represents the free propogator G0 and the vertices (dots) represent the interaction term. We then think of a given diagram in physical terms as the particle moving for a bit, then interacting with the external potential, then moving a bit, and so on. Quantum mechanics is a single-particle theory, so the diagramatic technique is not really needed. But in the standard model we shall see that it allows for a much simpler visualization of what is going on.

+ +

+ ...

(6.186)

Lecture seven - action functionals and Feynman rules

Aim: To introduce action functionals for the basic theories, and see how to read o the Feynman rules. Learning outcomes: At the end of this lecture you should know how to derive equations of motion from an action know how to write down the Feynman rules, once an action has been given.

7.1

action and equations of motion

A neat way of characterizing the dynamics of a system is through the action principle, which states that a system evolves in such a way as to extremize the action of a system. Such statements for static systems should be obvious by now, in that a static system has extremized its potential energy - a stationary ball on a slope must be either at the top of a hill, at the bottom of a valley, or at a saddle point. The action is the next step up, to include dynamical systems. This is not course on action principles, so we shall just proceed by example. 26

7.1.1

Klein-Gordon action

The action for the real Klein-Gordon scalar eld is given by S[] = 1 1 d4 x [ m2 2 ] 2 2 (7.187)

and it is this that we need to extremize. Now, when we nd the extrema of a function f (x) we solve the following equation
x0

lim f (x + x) f (x) = 0

(7.188)

which is just the rst-principles way of saying df /dx = 0. Here what we do is vary the eld , rather than x, and so we write
0

lim S[ + ] S[] = 0

(7.189)

In the case of the real KG eld we nd

lim

1 1 d4 x [ ( + ) ( + ) m2 ( + )2 ] 2 2

1 1 d4 x [ m2 2 ] = 0 2 2

We now expand this out, dropping the 2 as they are subdominant in the limit. We nd d4 x [ m2 ] = 0 Now we integrate by parts, dropping the boundary term, i.e. use = ( ) (7.191) (7.190)

and ignore the d4 x ( ) term. We shall always assume this vanishes; it amounts to saying there is no current at innity. Then d4 x [ m2 ] = 0 (7.192)

Now, we want this integral to vanish for all possible deformations , and the only way to achieve this is for the integrand itself to vanish, m2 = 0. (7.193)

This is just the Klein-Gordon equation (1.38). The point is that the action contains the same dynamical information as the eld equations, but it turns out that the action is easier to handle, especially when there are lots of eld. For example, the standard model contains 37 elds, so if we wanted to write it down in terms of equations of motion, that would be 37 rather complicated equations. However, in terms of the action, there is just one.

27

7.1.2

Dirac action

The Dirac action is given by S = d4 x[ + m] (7.194)

where = i 0 . This new object is because is a fermion, and so it needs this peculiar version of Hermitian conjugation. Now, despite all the complications one would expect from the Dirac equation, the variation of the action is remarkably simple. We treat and as independent objects, in which case the action must be extremized under variation of and separately. If we vary then we nd S = S[, + ] S[, ] = d4 x [ + m] = 0 (7.195)

and this must hold for all variations , in which case we must have + m = 0, (7.196)

which is just the Dirac equation (2.45). One can check that varying gives the same results, after a bit of work.

7.2

the structure of an action

The action, S, is usually written in terms of a Lagrangian density, L, where they are related by S = and L takes the form L = T V (7.198) d4 x L (7.197)

where T is the kinetic term, and V is the potential, which contins interaction term. There are a few useful things about the action that help to restrict the possible theories we may write down, they are L is real L is a Lorentz scalar - there are no free spacetime indices, they all come in pairs L is a scalar with respect to internal symmetries From the examples given it is now clear that pieces in the action that are quadratic in elds will produce terms in the eld equations that are linear. So, the interaction terms that will be the perturbation away from the linear (free) theory must be cubic (or higher) in the action, allowing us to write L = L0 VI 28 (7.199)

where VI is the interaction potential, and is cubic (or higher) in the elds. In terms of Feynman diagrams, it is the L0 that leads to the free propogator, and the VI that produces the vertices. The vertices are constructed rather simply by looking at each term in VI : if VI is cubic in the eld, then the vertex has three free-propogators meeting at a point;

if VI is quartic in the eld, then the vertex has four free-propogators meeting at a point; (7.201)

if VI is quadratic in one eld, but also quadratic in another, then the vertex has two freepropogators of each species. (7.202)

In fact, each term in VI will come with a parameter in front of it, for example in VI = 4 , in which case we would associate with the vertex, just as we would associate the propogator with a basic line. 7.2.1 example, real KG action I 1 1 1 L = m2 2 4 2 2 4 has VI = so the equation of motion for this theory is 1 4 4 (7.204) (7.203)

(7.200)

and the Feynman rules are

Note that we shall not be worrying about factors of 2, 1 ,... for the vertices, we only want the 4 parametric scale.

m2 = 3

(7.205)

1 p2 + m2

(7.206) (7.207)

29

7.2.2

example, real KG action II 1 1 1 L = m2 2 3 2 2 3 (7.208)

has VI = so the equation of motion for this theory is 1 3 3 (7.209)

and the Feynman rules are

7.2.3

example, complex KG action 1 L = m2 ( )2 2 (7.213)

m2 = 2

(7.210)

p2

1 + m2

(7.211) (7.212)

has

and the Feynman rules are

here we see something new, arrows on the lines. This is because there is a conserved quantity associated to the eld, due to the global U(1) symmetry. As we have seen before, such a U(1) symmetry is leads to an electric charge for the particle. When we have such a symmetry then, as charge cannot just dissapear, there must be the same amount of charge entering a vertex as leaving it, the arrows help us to keep track of where the charge is.

VI = 30

1 ( )2 2

(7.214)

1 p2 + m2

(7.215) (7.216)

7.3

example, Dirac particle coupled to photons

We have seen the Lagrangian for a basic Dirac particle, to couple in electromagnetism we simply replace D (4.132), L = D m = m + iq A so

which is something dierent to those we have seen before, as there are now two elds involved. This just means that the vertex has two fermion lines, and one photon i p + m p2 + m2 q (7.219)

Again, we put an arrow on the fermion line as there is a U(1) symmetry, and so charge is conserved.

7.4

complex KG equation coupled to photons

To couple the complex KG equation to electromagnetism we simply replace D , giving the Lagrangian L = (D )(D ) m2 = m2 + iqA ( ) + q 2 A A and then the Feynman rules are

7.5

Feynman rules for the electroweak sector of the standard model

31

(7.217)

VI = iq A

(7.218)

(7.220)

(7.221)

p2 q

1 + m2

(7.222)

(7.223)

q2

(7.224)

Now that we have a grasp of where the Feynman rules come from we shall write down the rules of one of the sectors of the standard model, the electrowek sector, i.e. the part that tells us

about the electromagnetic force and the weak nuclear force. The matter particles involved in this are the leptons e e and the gauge bosons (force carriers) , , , (7.225)

It is also believed to contain the Higgs scalar, but this has not been observed, yet. The basic set of diagrams are as follows e e (7.227)

e W

W+

e e

e W+

Z 0 , W , .

(7.226)

(7.228)

Z0

gw

(7.229)

gw

(7.230)

Z0

gw

(7.231)

32

Although we have only explicitly given the rules for the muon and the tau lepton doublets.

Z0

gw

(7.232)

Z0

gw

(7.233)

e e

doublet, they also hold for the

7.6

Fermis golden rule

Now that we have our Feynman rules we need to do something with them. Each diagram tells us about the amplitude for a process to happen, for example the amplitude for an electron to anihilate an anti-electron producing a photon is given by diagram 7.227, and has amplitude e. However, we are interested in physical quantities, such as rates of events, and so we actually need the probability, rather than the amplitude, and this is given by the modulus-square of the amplitude. There is one extra factor to take into account, the density of nal states , which gives us the number of states that are available to be occupied. The nal result is Fermis golden rule

where Tij is the probability amplitude. 7.6.1 ee scattering

e e This diagram shows the scattering of two electron by a photon. The diagram contains two vertices, each of strength e, and so the amplitude of the diagram is given parametrically by e2 . 2 1 so the rate of this process is Recall that the ne structure constant, = 4e0 c 137 2 (7.236)

(i j) =

|Tij |2

(7.234)

e2

(7.235)

33

As the diagram contiains two vertices we say it is a second-order process. Another diagram that contributes to electron-electron scattering is e e

e e but this is fourth-order (four vertices) and subleading so can be ignored at the rst approximation. 7.6.2 decay

The muon is the heavy version of the electron and, as it is heavy, is liable to decay. The leading order diagram responsible for the decay is e 1 2 MW

and here we also include the internal propogator of the W boson, which we have approximated as M12 , which is ne for low-momentum scattering. In fact, we can use this to give us an W expression for the Fermi coupling constant. Before the discovery of the W , and Z 0 it was known that a muon could decay into an electron and two neutrinos, so Fermi proposed an interaction term of the form VI (F ermi) GF (e )( ) e e (7.239)

which has the diagram

It was this GF that was measured rst (with a value of 105 (GeV )2 ), and we can see from our g2 electroweak diagram that GF Mw . 2
W

Lecture eight - group theory

W e e 34

e4

(7.237)

gw gw

(7.238)

GF

(7.240)

Aim: To introduce the concept of a group, and some important examples of groups. Learning outcomes: At the end of this lecture you should

 know the denition of a group know how to check whether a set, in combination with a composition law, forms a group know what denes Abelian groups, Lie groups and orthogonal groups.

8.1

denition of a group

A group is simply a set of objects, supplied with a rule that allows you to take two of these objects and make another one. This law of composition is denoted with a dot, g1 g2 = g3 , (8.241)

where gi are members of the set. For the set and composition law to be a group, G, it is required to satisfy the following it must be closed: If g1 , g2 G, then g1 g2 G it must have an identity, denoted e: e g = g e = g, g G. each g must have an inverse, denoted g 1 : g 1 g = g g 1 = e. the composition must be associative: g1 (g2 g3 ) = (g1 g2 ) g3 . Some examples should help to clarify this. Consider the set {1, i}, where the composition law is multiplication. This does not form a group because i i = 1, and -1 is not part of the set Consider the set {1, i, 1, i}, where the composition law is multiplication. This does form a group. Consider the set {1, i, 1, i}, where the composition law is addition. This does not form a group as, for example, 1 1 = 1 + 1 = 2 is not part of the set Consider the set of integers, Z, where the composition law is addition. This does form a group Consider the set of integers, Z, where the composition law is multiplication. This does not form a group as, for example, 2 does not have an inverse within the set, the inverse of 2 is 1/2, which is not an integer. The above examples should really be called representations of groups, which just means that they are explicit examples of the more abstract group itself. For example, we could say that the group we are interested in (called Z2 ) consists of elements {e, a}, such that a a = e. In this denition we make no comment about what a, e actually are, nor do we say what the composition law is. An explicit representation of this group could be {e 1, a 1, multiplication} but there are many others. We shall see more consequences of dierent representions later.

35

8.2

direct product of groups

In the standard model of particle physics there are a number of distinct groups, acting as independent internal symmetries, so we need a way of combining groups together, the direct product does just that. Suppose we have two groups G1 and G2 , we may form a group G3 = G1 G2 with elements g3 = (g1 , g2 ) and composition law g3 g3 = (g1 g1 , g2 g2 ), and it is simple to show that G3 is indeed a group. For example, consider the group D2 = Z2 Z2 , this is composed of elements {(e, e), (a, e), (e, a), (a, a)} where, for example (e, a) (a, a) = (e a, a a) = (a, e).

8.3

Abelian groups

Abelian groups are the simplest, they are the ones for which g1 g2 = g2 g1 . (8.242)

For example, the group formed by multiplication within the rational numbers is Abelian; addition is also Abelian. However, there are important examples that are not Abelian, with the non-Abelian nature coming typically because matrix multiplication is not commutative, AB = BA.

8.4

Lie groups (pronounced Lee groups)

The previous examples were all discrete groups, meaning that there was a countable number of elements in the set; the next step is to think about non-countable sets. For example the set of rational numbers, excluding zero, forms a group under multiplication (zero is excluded because it has no inverse under multiplication). More common examples are those associated to some sort of rotation. Consider the unitary group U(1), or rather its fundamental representation, dened by the elements g = ei (8.243)

with the composition law being multiplication (it is easily checked that this forms a group). This just corresponds to rotations in the complex plane by angle . and as is a continuous parameter this is a Lie group. 8.4.1 general linear group GL(n,R)

The fundamental representation of GL(n,R) is dened to consist of elements A where A is a real n n matrix det(A) = 0 The dimension of this group (the number of indepenedent numbers within an element) is n2 , as there are n rows and n columns.

36

8.5

special linear group SL(n,R)

The fundamental representation of SL(n,R) is dened to consist of elements A where A GL(n,R) det(A) = 1 The dimension of this group is n2 1, as there are n rows and n columns, and one real constraint coming from det(A) = 1.

8.6

orthogonal group O(n,R)

The fundamental representation of O(n,R) is dened to consist of elements A where A GL(n,R) AAT = I = AT A The dimension of this group is 1 n(n 1), which we now explain. Think of the matrix B, where 2 B = AT A. By constraining B = I we are requiring the n terms along its diagonal to be 1, i.e. n constaraints. The same constraint also sets all n2 n o-diagonal terms to zero, however, as B is symmetric this only imposes 1 (n2 n) independent constraints. The total number of degrees 2 1 of freedom is then n2 [n + 2 (n2 n)] = 1 n(n 1). 2 Note that AAT det(AAT ) det(A) det(AT ) det(A) det(A) det(A) = = = = = I 1 1 1 1 (8.244) (8.245) (8.246) (8.247) (8.248)

8.7

special orthogonal group SO(n,R)

The fundamental representation of SO(n,R) is dened to consist of elements A where A O(n,R) det(A) = 1
1 The dimension of this group is the same as that of O(N,R), 2 n(n1), because imposing the unit determinant only selects a sector of O(N,R) rather than reducing its dimension, as we already have det(A) = 1 for orthogonal matrices.

37

8.8

O(p,q,R)

The notion of orthogonal matrices is easily extended to O(p,q,R) by thinking of the dening relation of O(n,R) as AIAT = I Then we dene elements of O(p,q,R) by requiring AAT = (8.250) (8.249)

where diag(1, 1, 1, ..., 1, 1, 1...), with p minus ones, and q plus ones. This may seem like an obscure thing to do, but in fact this is just the Lorentz group of special relativity if we take p = 1, q = 3.

8.9

symplectic group Sp(2n,R)

Dene the symplectic matrix := 0 I I 0 (8.251)

where I is the n n identity matrix, then the symplectic group is dened to be those elements that satisfy A GL(n,R) AT A =

Lecture nine - non-Abelian gauge theory

Aim: To introduce the concept of a Lie algebras, and non-Abelian gauge theory Learning outcomes: At the end of this lecture you should know the denition of unitary groups know what group generators are, and how to calculate their basic properties know how to gauge a non-Abelian symmetry

9.1

unitary groups

Although the groups presented in the previous lecture all have their role to play in some area of physics, we are particularly interested in a dierent set of groups, the unitary groups, composed of complex matrices satisfying certain properties as follows

38

9.2

unitary group U(n)

We have already seen a unitary group, U(1), here we generalize that to matrices by saying U U(n) if U U = I = U U where U = (U T ) . (9.252)

The dimension of U(n) is n2 , which we see as follows. First note that there are n2 complex entries in U , corresponding to 2n2 real degrees of freedom. Now the diagonal part of U U = I 1 gives us n real constraints, while the o-diagonal part gives 2 2 (n2 1) real constraints (the 1 2 is because U U is symmetric and the 2 because the entries are complex). This gives a total of 2n2 [n + (n2 n)] = n2 real degrees of freedom for each U .

9.3

special unitary groups

U SU(n) if U U(n) det(U ) = 1 The dimension of U(n) is n2 1, because the determinant imposes one real constraint, which we see by considering U U det(U U ) det(U ) det(U ) |det(U )|2 det(U ) = = = = = I 1 1 1 ei (9.253) (9.254) (9.255) (9.256) (9.257)

so, if we impose det(U ) = 1, all we are doing is requiring the single real constraint = 0.

9.4

unitary groups and the standard model

In the standard model one nds that the paricles are described by their transformation properties under three dierent unitary groups, each with its own physical interpretation U(1) describes electromegnetism. It has dimension one, and gives us the single photon SU(2) describes the weak nuclear force. It has dimension three, and gives us the W and Z 0 bosons. SU(3) describes the strong nuclear force. It has dimension eight, and gives us the eight gluons. 39

9.5

generators and Lie algebras

We have already introduced the notion of generators when we were thinking about how to rotate a fermion eld; the concept of generators is also useful for internal symmetries. Simply stated, the generators X a and the group elements g of a Lie group G are related by g = ei
aXa

(9.258)

Once we have xed our choice of generators, the parameters a are what give us our dierent group elements. The reason that the generators are studied, rather than the group elements, is that for small a this just corresponds to looking at group elements near the identity, g = ei
aXa

I ia X a + ...

(9.259)

and that, it turns out, makes our life simpler. The generators X a form what is called a Lie algebra. 9.5.1 o(n), the Lie algebra of O(n)

The dening property of O(n) is enough to tell us that the generators of O(n) are skewsymmetric, we see this as follows OT O (I ia T a + ...)T (I ia T a + ...) a [T a + (T a )T ] Ta = = = = I I 0 (T a )T (9.260) (9.261) (9.262) (9.263)

Now, recall from the question sheet that O(n) is just the set of matrices associated with rotations (actually also reections), but we already have a set of generators for three-dimensional rotations given by (3.83), (3.85), which we note are indeed skew-symmetric. Although we chose these generators to do our calculations, we could have in fact chosen any set of linearly independent skew-symmetric matrices - it would just hve made the calculations harder and less intuitive. 9.5.2 su(n), the Lie algebra of SU(n)

Again, we may use the deneing property of SU(n) to tell us something about its generators. U U (I ia X a + ...) (I ia X a + ...) a [X a (X a ) ] Xa = = = = I I 0 (X a ) (9.264) (9.265) (9.266) (9.267)

So, the generators are Hermitian. However, we also need to impose unit determinant det(I ia X a ) = 1 1 ia T r(X a ) = 1 T r(X a ) = 0 40 (9.268) (9.269) (9.270)

so, as well as being Hermitian, the generators of SU(n) are traceless. Now, again, we already know of a set of matrices that are Hermitian and traceless - the Pauli matrices. And they do indeed generate SU(2). However, any set of traceless and Hermitian matrices would do, its just that Pauli matrices make life easier.

9.6

non-Abelian gauge theory

This is what the standard model is all about, and fortunately its not much dierent, conceptually, from electromagnetism - the maths is a little bit trickier though. To see how it works, consider the theory described by the Lagrangian L = m2 where is not just a complex scalar, but is a complex vector of scalars 1 2 = . . . n (9.271)

(9.272)

Now, given the denition of U(n), it is clear that this Lagrangian is invariant under U , for U SU(n), and where the matrix U does not depend on spacetime, i.e. it is a constant matrix. We can see this because if = U then = (U ) (U ) = U U = , = (U ) (U ) = U U = (9.273) (9.274)

Now we make the same grand statements as we did for U(1) electromagnetism, and say that we want this SU(n) symmetry to be local, and not just global. To do that we again introduce a covariant derivative D = iqA and require (D ) = U (x)D (9.276) (9.275)

when we promote the constant U matrix to a spacetime dependent matrix U (x). The only dierence is that now the vector potential A is actually a matrix, so we have to be a bit careful about how we manipulate expressions involving it, for example A A = A A and U A = A U . With the introduction of the gauge-covariant derivative D we may easily construct a gauge invariant Lagrangian L = (D ) (D ) m2 (9.277)

41

The next step is to determine how A changes under the gauge transformation, which we do with the following calculation (D ) iqA (U (x)) iqA U (x) U (x) + U (x) iqA U (x) U (x)D U (x)( iqA ) U (x) iqU (x)A U (x) iqU (x)A i A U (x) = U (x)A U (x) q i A = U (x)A U (x) [ U (x)]U (x) q = = = = (9.278) (9.279) (9.280) (9.281) (9.282) (9.283)

which joins = U to give us the full set of non-Abelian gauge trnsformations. We should compare this to the U(1) case (4.128, 4.129) to see that it does at least reduce to electromagnetism in the Abelian case.

10

Lecture ten - some representation theory

Aim: To reintroduce the idea of SU(2) representations Learning outcomes: At the end of this lecture you should be aware that there are many dierent guises for the same symmetry group know how to use ladder operators to construct the states transforming in a given representation of SU(2)

10.1

SU(2)

Because SU(2) is such an important group we shall spend a lecture going into some more of its properties. By now we have seen the algebra [j 1 , j 2 ] = ij 3 , +cyc. perm (10.284)

many times - this is the denition of the algebra of the generators of SU(2). At this point however, we have made no statement about what the j a are, nor what there composition law is. We have seen examples where the j a are matrices, and the composition is matrix multiplication, but they need not be matrices. Another important representation of this algebra comes from the dierential operators Lz = ixy + iyx , +cyc. perm (10.285)

which also satisfy this algebra. This should not come as a surprise, because these are just the orbital angular momentum operators. Even if the j a are matrices, there are an innite number of dierent matrix representations. the three-vector rotation generators (3.83,3.85) satisfy [T 1 , T 2 ] = iT 3 , 42 +cyc. perm (10.286)

and the Pauli matrices satisfy [ 1 /2, 2 /2] = i 3 /2, +cyc. perm (10.287)

And, as these two matrix representations consist of 2x2 and 3x3 matrices, they are not trivially related, and form distinct representations. In fact, we can continue, and nd a 4x4 representation 3/2 0 0 0 3/2 0 1 0 R1 = (10.288) 0 1 3/2 0 3/2 0 0 0 0 i 3/2 0 0 i 3/2 0 i 2 0 R = (10.289) 0 i 0 i 3/2 0 0 i 3/2 0 R3 = diag(3/2, 1/2, 1/2, 3/2) For later convenience we note that ( 1 /2)2 + ( 2 /2)2 + ( 3 /2)2 = (T 1 /2)2 + (T 2 /2)2 + (T 3 /2)2 (R1 /2)2 + (R2 /2)2 + (R3 /2)2 1 1 +1 I 2 2 = 1 (1 + 1) I 3 3 +1 I = 2 2 (10.291) (10.292) (10.293) (10.290)

This is all very interesting, but it would not be very convenient if we had to go through all of this for each representation, we need a way of unifying the calcuation. The way to acheive this is to nd a set of commuting operators, and use their eigenvalues to classify the representations. We have just seen that in the three cases we have looked at, (j 1 )2 + (j 2 )2 + (j 3 )2 is proportional to the identity, and so commutes with all the j a . In fact, one can show using the dening algebra that [(j 1 )2 + (j 2 )2 + (j 3 )2 , j a ] = 0. (10.294)

This gives us our rst commuting operator. The next one is a matter of choice, and its conventional to choose j 3 as the second commuting operator, giving us our commuting set of {j 2 = (j 1 )2 + (j 2 )2 + (j 3 )2 , j 3 } (10.295)

We are not saying that these commute with everything, only that everything in the set {j 2 , j 3 } commutes. For example, we could not add j 1 to the set, as it does not commute with j 3 . As is well known, for example from quantum mechanics, we are free to assign simultaneous eigenvalues to operators that commute, and that is what we shall now do. We dene j 3 to have eigenvalue m, and j 2 to have eigenvalue , and the eigenstate will be denoted |, m . j 3 |, m j 2 |, m = m|, m = |, m 43 (10.296) (10.297)

Now, rather than working with j 1 and j 2 directly, we introduce the raising and lowering operators j dened by j := j 1 ij 2 and note that the algebra gives us
2

(10.298)

[j 3 , j ] = j The reason for calling these raising/lowering operators is that j 3 [j |, m ] = = = = [j j 3 + (j 3 j j j 3 )]|, m [j j 3 j ]|, m j [j 3 1]|, m (m 1)[j |, m ]

(10.299)

(10.300) (10.301) (10.302) (10.303)

Lets explain what has just happened. We start with a state |, m that has a j 3 eigenvalue of m, then we showed that the state j |, m has a j 3 eigenvalue of m 1, i.e. j + raises the eigenvalue by one, and j lowers the eigenvalue by one. What this means is that we now have a way of constructing all the eigenstates, once one has been given Now, suppose we are given a value of , how should we go about constructing all the eigenstates? First of all we notice that if we choose the normalization , m|, m = 1 then = , m|j 2 |, m (10.304) (10.305) (10.306) = , m|(j 1 )2 + (j 2 )2 + (j 3 )2 |, m > m2

which tells us that there is a maximum, and minimum, value for m. So, at some value of m, usually denoted j, we will nd j + |, j = 0, because there are no more states with a higher eigenvalue than j by assumption. Now, a little algebra shows that j 2 = j j + + j 3 (j 3 + 1) so we nd j 2 |, j but j 2 |, j = |, j so = j(j + 1). (10.310) = [j j + + j 3 (j 3 + 1)]|, j = j(j + 1)|, j (10.308) (10.309) (10.307)

Just as j is the largest possible eigenvalue for j 3 (by assumption), j is the smallest eigenvalue. We see this by noting that j 2 = j + j + j 3 (j 3 1), so j 2 |, j
2

= [j + j + j 3 (j 3 1)]|, j = [j + j + j(j + 1)]|, j

(10.311) (10.312)

this technique has already been seen in the quantum mechanics of a harmonic oscillator, eg lec 20 of quantum World

44

but j 2 |, j = j(j + 1)|, j so j |, j = 0. (10.313)

To sum up, a given representation is specied by its j 2 eigenvalue, and a particular state within that representation is specied by its j 3 eigenvalue. As a matter of notation, it is more standard to denote states by |j, m . Now, what are the possible values for m? We know that there are an integer number of states, with j 3 eigenvalues ranging from j, j + 1, ...j 1, j, i.e. a total 2j + 1 states. As there is an integer number of states then 2j + 1 is an integer, then j must be either an integer, or half-integer. Here are some examples j = 0 rep: The only possible value for m is zero. An example of this is the s-state of the hydrogen atom, corresponding to l = 0 angular momentum.
1 j = 1 rep: The possible states for this are m = 2 , an example being the two spin-states 2 of an electron. We also note that this rep has j 2 = j(j + 1) = 1 ( 1 + 1), which matches 2 2 the Pauli matrices (10.291)

j = 1 rep: The possible states for this are m = 1, 0, 1, with an example being the l = 1 p-state of the hydrogen atom. Again, note that (10.292) has j 2 = j(j + 1) = 1(1 + 1), and so is a j = 1 rep. j=
3 2 1 3 rep: the possible states are m = 3 , 1 , 2 , 2 , an example being the Ra rep (10.293) 2 2

j = 2 rep: this has m = 2, 1, 0, 1, 2, and an example is the l = 2 d-state of the hydrogen atom. in fact, for orbital angular momentum we just make the identication j l, and note that l must be integer

10.2

normalizing the states

Although we have seen that j takes us from |j, m to |j, m 1 , we have to be careful of normalization. From our eigenvalue calculation, all we can actually say is that j |j, m |j, m 1 It is convenient to choose a normalization such that j, m|j, m for all m and j. So, we do this by |j, m + 1 = N j + |j, m (j ) = f j, m + 1|j, m + 1 = |N |2 j, m|j j + |j, m 1 = |N |2 j, m|j 2 j 3 (j 3 + 1)|j, m 1 = |N |2 [j(j + 1) m(m + 1)] j, m|j, m 1 = |N |2 [j(j + 1) m(m + 1)] 45 (10.316) (10.317) (10.318) (10.319) (10.320) = 1 (10.315) (10.314)

we are at liberty to choose the phase of N , and we pick it to be zero, so we end up with j + |j, m = j(j + 1) m(m + 1)|j, m + 1 (10.321)

if we perform a similar calculation starting with |j, m 1 = N j |j, m we nd j |j, m = j(j + 1) m(m 1)|j, m 1 (10.322)

10.3

general SU(2) represenation

it may interest you, but most probably not, to have a general expression for SU(2) generators of all dimensions 0 dl 0 dl 0 dl1 0 dl1 0 1 D1 = (10.323) , .. 2 . 0 dl+1 dl+1 0 0 dl 0 dl 0 dl1 0 i 0 dl1 D2 = (10.324) , .. 2 . 0 dl+1 dl+1 0 l l1 l2 D3 = (10.325) , .. . l + 1 l where dk = l(l + 1) k(k 1) (10.326)

11

Lecture eleven - isospin

Aim: To introduce the concept of isospin Learning outcomes: At the end of this lecture you should know how isospin relates to neutrons and protons be able to use isospin considerations in basic scattering calculations 46

11.1

neutrons, protons and isospin

At a very basic level, we note that the mass of a proton, p(938MeV), and neutron, n(940MeV), are very close. We also note that the excitation energy levels of mirror nuclei (where p and n are swapped) are very similar, e.g. 13 C and 13 N are mirror, as are 7 Li and 7 Be, and also 11 B 6 7 3 4 5 and 11 C. Heisenberg suggested that this points to the idea that neutrons and protons are, from 6 the perspective of nuclear dynamics, the same particle, or rather dierent states of the same particle. This is not such a wild idea, we do not think of the spin up and spin down states of an electron as separate particles. In fact, that analogy is rather apt, as the spin 1/2 property of an electron is going to become an isospin 1/2 property of the neutron/proton. So, we introduce the nucleon |N = a b (11.327)

with |a|2 + |b|2 = 1 to get the normalization, and identify |p So that |N along with the orthogonal state |N which we can combine into |N |N = a b b a |p |n =U |p |n (11.331) = b |p + a |n (11.330) = a|p + b|n (11.329) = 1 0 , |n = 0 1 (11.328)

where we see that U is in fact an SU(2) matrix. And that is it. The point about doing all this is to show that, once again, SU(2) is at the heart of things. The proton and neutron form a doublet, i.e. a j = 1 rep, of SU(2); and we call that SU(2) isospin, with the proton being the 2 1 m = + 2 state of isospin and the neutron being the m = 1 isospin state. 2 Because we have given this SU(2) a special name, it is also convential to use the notation j I, along with our standard notation for states 1 1 |p = | , , 2 2 1 1 |n = | , . 2 2 (11.333) m I 3. (11.332)

47

11.2

pions and isospin

So now we know what the I = 1 states of isospin are, what happens next? Well, from our 2 analysis of SU(2) in general, we know that there are an innitude of SU(2) representations, how do they make an appearnce here? For example, what do the I = 1 states correspond to? Well, the rst thing we know is that there should be three particles in this multiplet, corresponding to I 3 = 1, 0, 1, and that they should all have roughly the same mass - as they are all related by an SU(2) transformation. Fortunately such a set of particles exists, the pions, + (140MeV), 0 (135MeV), (140MeV), and we identify these with the I 3 = 1, 0, 1 states of the I = 1 isospin representation. The isospin system does more for us than just classify particles, it tells us about the isospin composition of the elements within a multiplet. This concept will be of great value when we come to quarks, so we shall take a close look at how, from the isospin perspective, the pion triplet is constructed from the nucleon doublet. Given the two states |p and |n there are four independent combinations of their product that we may construct: |p |p , |p |n , |n |p , |n |n , and these may be arranged into the symmetric states |p |p , 1 (|p |n + |n |p ), 2 |n |n , I=1 (11.334)

to form the I = 1 triplet of pions, and 1 (|p |n |n |p ), 2 I=0 (11.335)

to form the I = 0 singlet. From the point of view of group theory, what we have shown is that 22 = 13 (11.336)

i.e. the direct product of two doublets gives us an SU(2) triplet and an SU(2) singet. It may not be obvious why we should look at the symmetric and anti-symmetric decompositions, but it is easily checked that they are the correct ones by using the generators on the product space
3 Iprod = I 3 I + I I 3 Iprod = I I + I I

(11.337) (11.338)

where the product of generators acts in the natural way, for example (I 3 I + I I 3 )|p |p = (I 3 I)|p |p + (I I 3 )|p |p = (I 3 |p ) (I|p ) + (I|p ) (I 3 |p ) 1 1 = ( |p ) (|p ) + (|p ) ( |p ) 2 2 1 = (|p |p + |p |p ) 2 = 1 |p |p 48 (11.339) (11.340) (11.341) (11.342) (11.343)

i.e. the state |p |p has eigenvalue I 3 = 1, which is what we claimed. All of this may be stated as follows for the pion triplet | + = |1, 1 | 0 = |1, 0 | = |1, 1 1 1 1 1 = | , | , = |p |p 2 2 2 2 1 1 1 1 1 1 1 1 1 = | , | , + | , | , 2 2 2 2 2 2 2 2 2 1 1 1 1 = | , | , = |n |n 2 2 2 2 (11.344) 1 = (|p |n + |n |p )(11.345) 2 (11.346) (11.347) so, we have the nice results that the pions are constructed out of nucleons, but we get an extra state, the singlet. This is associated with the deuteron, |d , coming from the anti-symmetric product of nucleon states. |d = |0, 0 1 = 2 1 1 1 1 1 1 1 1 | , | , | , | , 2 2 2 2 2 2 2 2 1 = (|p |n |n |p ) (11.348) 2 (11.349)

Or, equivalently |p |p |p |n |n |p |n |n = | + 1 = | 0 + |d 2 1 = | 0 |d 2 = | (11.350) (11.351) (11.352) (11.353)

11.3

using isospin in scattering calculations

This is all very nice, but it is actually also quite useful, as it tells us something about the scattering of particles. If isopspin really is a useful quantum number, then it should be conserved, so we can use this to rule out, or constrain certain scattering rates. For example, lets consider the two scattering processes pp d + , The amplitude of a scattering event is given by amplitude(initial f inal) = f inal|Hint |initial so we need the bras and kets associated with pp, pn, d + , d 0 . pp |p |p pn |p |n d + |d | + d 0 |d | 0 1 1 1 1 =| , | , 2 2 2 2 1 1 1 1 = | , | , 2 2 2 2 = |0, 0 |1, 1 = |0, 0 |1, 0 49 = |1, 1 , 1 = (|1, 0 + |0, 0 ) , 2 = |1, 1 = |1, 0 (11.356) (11.357) (11.358) (11.359) (11.355) pn d 0 (11.354)

Now, given that isospin is conserved, it commutes with the Hamiltonian, meaning that the initial and nal states must have the same isospin eigenvalues, or else the amplitude will vanish. It is then immediately clear, for example, that we cannot have pp d 0 , because the pp state and the d 0 state have dierent isospin eigenvalues. We can, however, have pn d 0 , because the pn state contains a bit of |1, 0 . In fact, we can make a prediction for the ratio of cross sections as follows amp(pn d 0 ) = amp(pp d + )
1 2

( 1, 0| + 0, 0|) Hint |1, 0 1, 1|Hint |1, 1 0, 0|Hint |1, 0 1, 0|Hint |1, 0 + 1, 1|Hint |1, 1 1, 1|Hint |1, 1 1, 0|Hint |1, 0

(11.360) (11.361) (11.362) (11.363)

1 = 2

1 = 2 1, 1|Hint |1, 1 1 = 2

the last step may not be obvious, but recall that (10.321) |1, 1 I I + 1 = I + |1, 0 2 2 = I I 3 (I 3 + 1). (11.364) (11.365)

Now that we have the ratio of the amplitudes, we nd the ratio of cross sections by squaring, giving (pn d 0 ) 1 = , +) (pp d 2 which matches observations. (11.366)

12

Lecture twelve - the quark model

Aim: To introduce where quarks came from Learning outcomes: At the end of this lecture you should know how to construct the baryon octet and decuplet know the particles in the baryon octet and decuplet, and their quark content be able to state the Gell-Mann Nishijima formula

12.1

more multiplets
3 2

The success of isospin continued with the identication of the I = of particles ++ (1230M eV ), + (1231M eV ), 0 (1232M eV ), 50

quadruplet as the -series (12.367)

(1232M eV ).

and it was noted that the known multiplets (nucleons, pions, deltas) sased 1 Q = I 3 + B, 2 (12.368)

where Q is the electric charge, and B the baryon number. e.g. has I 3 = 1, B = 0, Q = 1. This was all very nice, until more particles came along. The troublesome particle was one called te lambda, , which appeared without any charged partners, so was an isospin singlet with I = 0. That is ne, but it decayed via p (12.369)

and the state on the right hand side has non-vanishing isospin (lec 11 questions). Actually it was not not a total disaster, because the decay time was 1010 s, compared with the expected timescale for strong interactions of 1023 s. So, although it was produced by the strong interaction, it did not decay by it. Nevertheless, as it could be produced by the strong interaction in p collisions it would have to be produced in pairs, so that the collision could conserve isospin. This meant that along with the we also had another particle, the K 0 . In fact there were a host of pair-produced particles + K 0 K 0 p+ (12.370) 0 K Gell-Mann and Nishijima independently suggested that the new data could be explained by introducing a new quantum number, preserved by the strong interaction, strangeness, and assigned S(K + ) = 1, S( ) = 1. It was discovered that the lowest mass spin- 1 baryons tted neatly into the following octet 2 S=0 S = 1 + (1189M eV ) p(938M eV ) 0 (1192M eV ) (1116M eV ) 0 (1315M eV ) I 3 = 1/2 (940M eV ) I 3 = 1/2 n(940M eV ) (1197M eV )

S = 2 I3 = 1

I3 = 0

I 3 = 3/2

This new quantum number extended (12.368) to the Gell-Mann Nishijima formula 1 Q = I 3 + (B + S). 2 (12.371)

Along with the spin- 1 it was also discovered that the lowest-mass spin- 3 baryons t nicely into 2 2

51

the following decuplet S=0 S = 1 S = 2 S = 3 ++ +

+
0

I 3 = 3/2 I 3 = 1 I 3 = 1/2 I 3 = 0 I 3 = 1/2 I 3 = 1 I 3 = 3/2

In fact, the was actually a prediction, but they were right. Gell-Mann has subsequently commented that the next step was obvious, but he believed it to be more a mathematical trick, than a physical picture. The observation is that triangles are 1 the base shape of both the hexagonal pattern for spin- 2 baryons, and the triangular pattern of the spin- 1 baryons; so he introduced three quarks3 . He gave them the labels u, d, c, standing 2 for up, down and strange. Just as |n and |p were identied as the dierent states of the same particle, with an SU(2) symmetry relating them, so |u , |d , |s were considered the same, but with an SU(3) symmetry relating them, called avour; Baryons were then formed of three such quarks. We are then led to ask what we can form out of three quarks, and there are 27 possibilities: uuu,uud,udu,... However, rather than just list them in the basic form we should think bck to how our SU(2) isospin multiplets worked, for example Qu 5 of lecture 11. The multiplets form out of particular combinations of states that have symmetry under the interchange of pairs totally anti-symmetric, f lavour (1): uds + dsu + sud usd sdu dus anti-symmetric on the rst pair, f lavour (8)12 : (ud du)q, (us su)q, (ds sd)q where q is one of u, d, s. anti-symmetric on the second pair, f lavour (8)23 : q(ud du), q(us su), q(ds sd) where q is one of u, d, s. totally symmetric, f lavour (10): uuu, ddd, sss, uud + udu + duu, uus + usu + suu, ddu + dud + udd, (12.375) dds + dsd + sdd, ssu + sus + uss, ssd + sds + dss
3

(12.372)

(12.373)

(12.374)

Zweig had a similar idea, calling them aces instead.

52

On the face of it, this gives us too many states, 29 instead of 27; however, we have over-counted. A linear combination of the states in the list that are anti-symmetric in the rst pair is also toally anti-symmetric, but that one has already been counted, the same goes for the list of states that are anti-symmetric in the second pair. So, we nd the nal result 3 3 3 = 1 8 8 10 (12.376)

and that is more like what we need, explaining the origin of the decuplet and the octet of baryons. However, we seem to have an embarrassment of riches, as there appears to be an extra octet and singlet that are not observed. To explain these we need to remember that 1 quarks are spin 2 particles, and baryons are fermions so their overall wavefunction needs to be anti-symmetric. The total wavefunction consists of total = space spin colour f lavour (12.377)

and we know that physical states are colourless so the colour wavefunction is the totally antisymmetric colour singlet. For lowest enery states we have no angular momentum (l = 0), which is symmetric, so we are left with requiring spin f lavour to be totally symmetric. We calculated in question 5 of lecture 11 that the product of three spin- 1 states gave 2 2 2 = 4 2 2 2 totally anti symmetric 1 anti symmetric on f irst pair, spin ( )12 2 1 anti symmetric on second pair, spin ( )23 2 3 totally symmetric, spin ( ) 2 : none

: ( ) , ( ) : ( ), ( ) : , + + , + + ,

which means that we may form a totally symmetric spin f lavour out of 3 f lavour (10) spin ( ) 2 (12.378)

as each of the components are totally symmetric, or we can form the totally symmetric state 1 1 1 f lavour (8)12 spin ( )12 + f lavour (8)23 spin ( )23 + f lavour (8)31 spin ( )31 2 2 2 (12.379)

and that is all. So, the extra singlet and octet are in fact not allowed because the quarks are spin- 1 . Moreover, this explains why the baryon decuplet is spin- 3 (it is made out of the spin- 3 2 2 2 part of 2 2 2, i.e. the 4 in 2 2 2 = 4 2 2) and the baryon octet is spin- 1 (it is made 2 1 out of the spin- 2 part of 2 2 2, i.e. the 2s in 2 2 2 = 4 2 2). Since the early days the quark model has been extended, as more particles were discovered due to higher energy colliders coming online, we are now at the stage where the quark model is given as in table 1 and the Gell-Mann Nishijima formula has extended to 1 Q = I 3 + (B + S + C + B + T ). 2 (12.380)

53

13

Lecture thirteen - charge conjugation and parity

Aim: To introduce discrete symmetries Learning outcomes: At the end of this lecture you should know how to nd the parity of photons and mesons know the parity of fermions, anti-fermions and fermion-antifermion pairs be able to argue that electric dipole moments of particles would violate parity

13.1

Charge conjugation

We have seen that the standard model relies on continuous Lie symmetries to give us vector bosons, such as photons and gluons, but then we also need to think about discrete symmetries. For example, does the Universe care about a distinction between left and right - can we ip the Universe in a mirror, and leave it looking the same? Can we swap all particles in the Universe with their anti-particles without anyone noticing? The second of these is known as charge conjugation, C, and it acts on particle states as C|p = | p (13.381)

where | is the anti-particle of |p . We can nd the eigenvalues of C by noting that C2 = 1 p CC|p = C| = |p p (13.382)

so C has eigenvalues 1. We also see that only particle that are there own anti-particles can be C eigenstates, i.e. the eigenvalue equation C|p = |p implies that | = |p . p quark Q B I I3 S C d -1/3 1/3 1/2 -1/2 0 0 u 2/3 1/3 1/2 1/2 0 0 s -1/3 1/3 0 0 -1 0 c 2/3 1/3 0 0 0 1 b -1/3 1/3 0 0 0 0 t 2/3 1/3 0 0 0 0 B 0 0 0 0 -1 0 T 0 0 0 0 0 1 mass (MeV) 5 2 100 1,200 4,200 174,000 (13.383)

Table 1: Table showing quark charges

54

13.1.1

photons

Recal that the photon is just the quantum version of electromagnetism, and that one, actually two, of Maxwells equations are F = 0 j Now, it is clear that under charge conjugation the sign of the current will change C : j j so to maintain Maxwells equations we better have (13.385) (13.384)

Showing that the photon has negative intrinsic parity, and that electromagnetism preserves C. 13.1.2 fermion-antifermion pairs

Consider the decay of a lowest energy state scalar meson, i.e. l = s = 0, into two photons As electromagnetism preserves C, then the initial and nal states must have

Now suppose we consider the spin-ip of a meson, i.e. an s = 1 meson decays to an s = 0 meson plus a photon. As the photon has C = 1, and the s = 0 meson has C = 1, then the s = 1 meson must have C = (1)(1) = 1 = (1)s Now suppose we have a meson in a zero spin, but non-zero angular momentum state, l, decaying to an l 1 angular state via the emmision of a photon, the same argument as above shows that it has C = (1)l . We combine all the above results into the single relation that C|f f , l, s = (1)l+s |f f , l, s (13.387)

C : A A

(13.386)

the same C value, and we know that the photon has C = 1, so a nal state of two photons has C = (1)(1) = 1, meaning that the l = s = 0 meson has C = 1.

This has important consequences for particle decays, and explains why certain decays do not occurr, for example the scalar meson 0 in the l = 0 state has C = 1 so 0 0 is allowed is not allowed (13.388) (13.389)

55

13.1.3

C violation

Unfortunately, or fortunately depending on your point of view, the charge conjugation symmetry is not a symmetry of the full standard model. C is only a symmetry of the strong sector and the electromagnetic sector, with the weak nuclear force breaking charge conjugation symmetry. For example, the following decay is observed + + (L) (L) (13.390)

with the muons always coming out left-handed. If we now perform charge conjugation on both sides we get (L) (L) (13.391)

which looks ne, until you realize that experimentally all the muons from decay actually come out right-handed, so this weak sector decay (13.390) violates C.

13.2

parity

The question of whether the Universe looks the same in a mirror is answered by considering parity symmetry, where we take the co-ordinates x x. When we do this we need to know how physical quantities change, for example the momentum p p, and so is an example of a polar vector - one that changes sign along with the co-ordinates. Angular momentum, however, changes according to L L, because L = r p; this is an example of a pseudo vector. This behaviour is not limited to vectors, scalars also split up into scalars and pseudo scalars. Electric charge is an example of a scalar as , whereas the volume of a parallelipiped dened by vectors a, b and c, a (b c), is a pseudo scalar, changing sign under parity. 13.2.1 electromagnetism

Electromagnetism is invariant under parity, and this can be seen by assigning the photon negative intrinsic parity as follows. From Maxwells equation E = /
0

(13.392) requires (13.393)

we see that invariance of this equation under P : , P : P : E E. However, E = showing that P : , A A. A t

(13.394)

(13.395)

56

Now recall that the four-vector potential A = (, A) and we see that P : A A so we associate a photon state with negative intrinsic parity P | = | (13.397) (13.396)

Now that we know how the vector-potential behaves we can see that the magnetic eld is a pseudo vector, as B = 13.2.2 fermions A. (13.398)

Fermions are a little trickier to gure out, but the result of quantum eld theory is that fermions and anti-fermions satisfy P |f = |f , P |f = |f (13.399)

so fermion and anti-fermion have opposite parity. In particular, the s-state (vanishing angular momentum) of the scalar meson 0 has negative intrinsic parity, so is actually a pseudo scalar meson. But now we seem to have a puzzle. If a meson has negative intrinsic parity, then why is the following allowed 0 (13.400)

as the nal state, which consists of two photons, appears to have positive parity. The point is that the parity must take into account the wave function, as well as the intrinsic parity, so if the wavefunction of the two-photon state has negative parity, this leaves a negative parity state. For example, write a wavefunction in terms of spherical harmonics (x) = (r)Ylm (, ) where parity corresponds to x x then P : Ylm (, ) (1)l Ylm (, ). (13.403) , + . (13.402) (13.401)

For example, Y11 sin ei sin (ei ). So, if the two photons are in an l = 1 state, then the overall parity is negative. Another way to think about it is to ask how the pions couple to the electromagnetic eld. We know from our Feynman diagram work that if we want a single pion to decay to two photons then the Lagrangian needs to have a term containing a single pion eld, and two photons, and that such a term must be a scalar. There are two choices L1int 0 E B L2int 0 (E E + BB) 57 (13.404) (13.405)

In the rst choice, as E is a vector and B is a pseudo vector, we must have the pion as a pseudo scalar for L1int to be a scalar. The second choice leads to the pion being a scalar, which is not what is observed. In fact, the E B coupling actually tells us that the photons come out of the experiment with orthogonal polarizations, which is how it was gured out experimentally. All of this argument about including the spatial wavefunction into the parity assignment goes through for the fermion-antifermion pair, so in fact the parity for a meson state is P |f f = (1)l+1 |f f (13.406)

13.3

linearity of parity

We now do something that may seem strange, which is to calculate the parity of i, the square root of -1. Although it will give us what we think of as an obvious result, it is worth doing as we will see that things change when we look at time reversal. So, lets start with the canonical commutator [, p] = i x (13.407)

Now we perform a parity transformation, remembering that operators transform as P xP 1 if states transform as P | . P xP 1 = , x to nd P [, p] P 1 = i x [, p] = P i P 1 x so for the commutation law to be preserved we must have P i = iP P iP 1 = i (13.411) (13.409) (13.410) P pP 1 = p (13.408)

which is just the statement that the parity transformation acts linearly. As we usually think of i as a number this result is what we would normally write down without thinking, so thats good, but we need to take care later on when we consider time reversal.

13.4

electron dipole moment (edm)

We now think of ways that parity could be violated, or rather how we could observe its violation. Suppose that a fundamental particle, which has a non-zero spin, had an electric dipole moment, this would violate parity symmetry. There are now two possibilities, the electric dipole moment could either align or anti-align with the spin; we shall go through the argument for the aligned case, but the same holds for anti-aligned. So, we place such a particle in a state where the spin (and electric dipole moment) both point up, now perform a parity transformation. The parity transformed state would have the spin pointing up (spin is a pseudo vector so doesnt change) and electric dipole moment pointing down. However, such a particle is not in our theory, we said 58

the particle had aligned spin and electric diole moment. So, the presence of an e.d.m. implies the violation of parity. For example, the edm of a neutrons and electrons are bounded by dneutron < 2.9 1026 e cm dneutron < 1.6 1027 e cm (13.412) (13.413)

which, to put into perspective, compares to the size of a neutron of around 1013 cm, i.e. dneutron < 2.91013 e lneutron . So, imagine the neutron is the size of the Earth, then 1013 e learth is 0.01e mm, corresponding to displacing an electron by 0.01mm in the centre of the Earth!

14

Lecture fourteen - parity violation and time reversal

Aim: To introduce discrete symmetries, and their breaking Learning outcomes: At the end of this lecture you should know how parity was found not to be a symmetry know what time reversal symmetry is

14.1

parity violation

In experiments that observed the decays of cosmic rays it was noted that two particles, which were otherwise identical, had two dierent decay channels, these were called the and + + 0 +00 + + (14.414) (14.415)

If these were indeed distinct particles then this would be ne, but if not, theres a problem with parity conservation. As pions have negative intrinsic parity then the parity of the + nal-state is positive, implying that + has positive parity, but the parity of the + nal-state is negative, implying the + parity is negative. For these to be the same particle (now caled K+ ), then the decays would have to violate parity - nobody wanted that. In 1956 Lee and Yang noted that there was an absence of data showing that parity ws conserved in the weak interaction, in contradistinction to data for the strong nuclear force and the electromagnetic force. In the same year Wu performed an experiment looking at the following decay of cobalt to nickel
60

Co

60

N i e e

(14.416)

Wu and collaborators aligned the spins of the cobalt nuclei by utilising their magnetic moment, which point in the same direction as their spin. The experiment revealed that most of the electrons were emitted in a direction opposite to the spin vector of the nucleus, a result which means that parity is not a symmetry in these decays. To understand that, think about the correlation between the direction of nuclear spin, and the momentum of the electrons, Corr = 59

s p . If parity is a symmetry then Corr should have the same value before and after a parity transformation, however, P : Corr Corr as s is a pseudo vector, and momentum a polar vector. Another way to see this is to imagine aligning the nuclear spins pointing up, then the experiment says that the electrons are mostly emmitted downwards. If we perform a parity transformation the spin still points up, but now the electrons have reversed their momentum, and are also moving upwards. So, the actual experiment and its parity-reversed version are dierent, so explicitly violating parity4 . Another place that parity is seen to be broken by the weak sector is in the decay of pions to muons. + + ,L L (14.417)

where it is observed that the muon always comes out left-handed. If parity were a good symmetry of this decay then we would expect + + ,R , as this is the parity reversed version of the R observed decay. Similarly, we observe ,R R (14.418)

but not its parity-reversed partner. The underlying reason for this is that all neutrinos are left-handed, and all anti-neutrinos are righthanded - thats about as parity violating as you can get!

14.2

time reversal symmetry

Although it may appear that at rst-sight that nature does not have a time-reversal symmetry, shattered glass never recombines to become whole again, this is really just because of entropy, rather then an asymmetry of the fundamental interactions. If we think about Newtons laws there is no problem in taking t t. Now we want to ask whether quantum mechanics has anything to say about it. Suppose that we have some map that takes un-primed states | to primed states | , for all states. Then Wigner showed that if the inner product is preserved | | |2 = | | |2 we must have either | = U | , or | = A| (14.420) (14.419)

where U is a unitary and linear operator, and A is anti-linear and anti-unitary. The terms unitary and so on are dened by U [| + | ] A[| + | ] U |U A|A
4

= U | + U | , = A| + A| , = | , = | = | ,

linear anti linear unitary anti unitary

(14.421) (14.422) (14.423) (14.424)

note that a lot of texts get this the wrong way around, using arguments like the spin changes direction in a mirror - be careful.

60

So, the unusual thing about an anti-linear operator is that when you move it past a complex number, you have to complex-conjugate it, i.e. they dont commute with numbers! To see why this makes an appearance in quantum mechanics consider again the canonical commutator [, p] = i x and now we perform a time-reversal transformation with operator T , T xT 1 = x, T pT 1 = p (14.426) (14.425)

so, whereas parity changes the sign of both x and p, time reversal changes only the momentum, meaning that x T [, p] T 1 = T i T 1 [, p] = T i T 1 x (14.427) (14.428)

So, if we are to maintain a time reversal symmetry in quantum mechanics, i.e. keep the commutation relations, then we must have T iT 1 = i, T i = iT (14.429)

so the time reversal operator is anti-linear. This result also manifests itself in the behaviour of wavefunctions under time reversal, consider for example te plane wave T : = ei(Etpx) ei(Etpx) = (14.430)

because t t and p p. So, we see that eectively i i. Under time reversal the spin changes sign, but the electric dipole moment does not so, just as a non-zero edm signals a breakdown of parity symmetry, it also implies a violation if time-reversal invariance. T may also be checked using a principle called detailed balance, for example, the scatterings np D, and D np (14.431)

should occur with the same rate if T is a symmetry.

15

Lecture fteen - CP symmetry

Aim: To introduce a product discrete symmetries, and its breaking Learning outcomes: At the end of this lecture you should recognise how the electroweak sector (almost) preserves CP know how C and P act on the neutral kaon system be able to calculate the CP violation in te neutral kaon system. 61

15.1

CP symmetry

We have seen that the weak sector violates both C and P individually, but there is still the possibility that together they are preserved. For example, the observed decay + + ,L L becomes the following under C and P and the product CP + + ,L + + ,R L R
C C P

(15.432)

,L ,R L R And in fact the CP (or PC) transformed decay is observed, so maybe the weak sector preserves CP even though each discrete symmetry is broken. In fact, it was argued by Landau that CP, rather than P, is the correct version of the discrete symmetry associated to mirror reections (see the questions associated with this lecture), and if we take that view, then maybe the Universe is mirror-symmetric afterall.

15.2

neutral kaons

A good laboratory for studying CP symmetry is the neutral kaon system, it is interesting because the the K 0 is able transform to its anti-particle, K 0 . This is achieved through the following processes, d s d s u W W , W u u (15.433)

s s d d where we start with a K 0 (a d bound state), and evolve to a K 0 . What this eectively means s is that from the perspective of the weak interaction the K 0 and K 0 are sort of the sme particle. However, as far as the strong force is concerned, they are denately dierent, for example K 0p K +n (15.434)

u W P |K 0 P |K 0 C|K 0 C|K 0 = = = = 62 |K 0 |K 0 |K 0 |K 0

is allowed through the strong force as it conserves strangeness, however, the strong force does not allow K 0 p K + n, as this would require a change in strangeness.

15.3

kaons and C, P and CP

as kaons are mesons (quark-antiquark pairs) then they have negative intrinsic parity, so we have (15.435) (15.436) (15.437) (15.438)

the sign change in the charge-conjugation operation is just convention. In particular, we note that kaons are not CP eigenstates CP |K 0 CP |K 0 = |K 0 = |K 0 (15.439) (15.440)

We now consider the view that the weak interaction preserves CP and see if that is consistent with experiment. So, the picture is that kaons are produced by the strong force (so are eigenstates of strangeness), and then they decay via the weak force, which by assumption means they will be CP eigenstates. Although the kaons themselves are not CP eigenstates, it is easy to construct them, we call the CP eigenstates |K1 and |K2 , |K1 |K2 so CP |K1 CP |K2 = |K1 = |K2 (15.443) (15.444) 1 = (|K 0 + |K 0 ) 2 1 = (|K 0 |K 0 ) 2 (15.441) (15.442)

These are the quantum states that are relevant for the weak decay of kaons, assuming the weak force preserves CP. Kaons decay mostly into pions, and CP conservation implies K1 K2 (15.445) (15.446)

as pions have negative intrinsic parity. This is useful because the decay into three particles is slower (0.5 107 s) than into two pions (0.9 1010 s). So, Cronin, Fitch and collaborators decided to set up a beam of kaons and watch what they decayed to. Given that the K1 decays so much faster than K2 , if we have a long enogh beam, then all of the k1 will have decayed by the end of the beam-pipe. So, we should expect only decays at the end of the beam - if CP is conserved in the weak sector. However, experiment showed that about 1 in 500 of the kaon decays at the end of the pipe was into two pions, direct evidence of CP violation in the weak sector.

15.4

Klong and Kshort decays

So, now that we know CP is violated, it does not make sense to analyse the system in terms of weak eigenstates, but rather we should use mass (or lifetime) eigenstates. Given the smallness of CP violation however, we know that the mass eigenstates are close to the CP eigenstates, so we write |KS |KL = = 1 1 + | |2 1 1+| |2 (|K1 |K2 ) (|K2 + |K1 ) (15.447) (15.448)

63

Now we look for decays of KL into nal states that are CP-conjugates of each other, for example the decays KL e+ e , KL + e e , rate 1 rate 2 (15.449) (15.450)

are allowed and, as we shall now see, allow us to distinguish between matter, and antimatter. 1 : K 0 can decay to e+ e , but not to + e e , via e e (15.451) u d e e (15.452) u

d 0 can decay to + e e , but not to e+ e , via 2 : K

d d What all this means is that the rate 1 is controlled by the amount of K 0 in KL , and the rate 2 is controlled by the amount of K 0 in KL , but we know that |KL so if we dene := to distinguish between the rates we nd = experiments give 103 , so 1 > 2 (15.456) |1 + |2 |1 |2 2 Re( ) |1 + |2 + |1 |2 (15.455) 1 2 1 + 2 (15.454) 1 = 2 1 1 + | |2 (1 + )|K 0 (1 )|K 0 (15.453)

W W 64

15.5

matter vs anti-matter

We now have a way of distinguishing matter from anti-matter. This is really quite remarkable, it would seem entirely natural to assume that our Universe could have been made from anti-matter rather than matter, without making any dierence, but this is not the case - an anti-matter Universe is dierent to a matter Universe. The way we can dene anti-matter uniquely is to say that it is the stu associated with the more common lighter charged product of KL decay, i.e. e+ . In fact, that there is an anti-symmetry between matter and anti-matter is a good thing for us, otherwise we would expect our Universe to be half matter and half anti-matter, in fact we observe a baryon to photon ratio of bb n b 6 1010 n compared to around 1040 that we would expect simply from statistical uctuations. (15.457)

16

Lecture sixteen - quark mixing

Aim: To extend the idea of state mixing to quarks Learning outcomes: At the end of this lecture you should know about the three generations of the standard model know how the Cabbibo mechanism allows for generation-changing processes be able to estimate the ratio of rates for certain basic processes.

16.1

three generations

The lepton sector of the standard model neatly splits into three generations, e e , , .

and in terms of Feynman diagrams the basic interactions occur with strength gw e u gw W gw W gw .

e d In particular, there is no weak decay that allows an electron to turn into a W and a , weak interactions do not mix the lepton generations. The next observation is that the quark sector also has three generations u d , c s 65 , t b . (16.460)

(16.458)

W (16.459)

which makes us think there could be a correspondence, i.e. does the e W e , with strength gw , diagram imply a d W u diagram of strength gw ? Moreover, does the non-existence of weak-sector generation process imply the impossibility of d W c? It turns out that this intuition is close, but not quite right. For example, along with W which just involves u, d quarks, we also nd K W which shows that the W must couple to both u and s, as K is a su meson. (16.462) (16.461)

16.2

quark mixing

We have already seen the importance of mixing quantum states in particle physics, this was how we explained the neutral kaon decays; here we just apply that notion to the quark sector. The point is that the weak and the strong forces are distinct, the interaction eigenstates of one need not correspond to the interaction eigenstates of the other. The correct quark weak-interaction eigenstates are actually a linear combination of the strong-interaction eigenstates, i.e. the weak nuclear force sees d , s , t rather than d, s and t. To keep things simpler, lets just think about the rst two generations, in which case one has

where c is the Cabbibo angle, i.e. they are related by a unitary matrix. The correct Feynman diagrams are then c u gw W gw W (16.464)

d s We may now, for example, test this idea through measurements of rates such as (K W ) W ) (

these rates involves the diagrams u W

d sin c + s cos c d cos c s sin c The point being that the K is a particle in an eigen state of the strong interaction, K = u = u(d sin c + s cos c ), and the pion is also a strong-force eigenstate, = d = (d cos c s u s sin c ). Now, in the rst diagram there is an us W vertex, but there is no such coupling, so u 66

d s = cos c sin c sin c cos c d s u W

(16.463)

(16.465)

(16.466)

the rst diagram contributes gw sin c to the amplitude, as there is a ud W vertex in the standard model and it has strength gw . In the second diagram the uW s vertex gives no contribution, meaning that the whole diagram contributes gw cos c to the amplitude. So,
2 (K W ) (gw sin c )2 = ( W ) (gw cos c )2 = tan2 c

(16.467) (16.468)

and experiments show that

Another way of stating this is to say the weak coupling strengths for quarks are gud = gcs = gw cos c gus = gcd = gw sin c (16.470) (16.471)

Now that we understand the coupling of quarks tothe W boson, we use the weak interactions e e gz Z gz Z (16.472)

e to predict the strong interactions d gz

d i.e. an interaction Lagrangian of the form

and therefore Feynman diagrams of the form d gz

d There are no neutral weak interactions that change, for example, strangeness.

17

Lecture seventeen - neutrino oscillations

c 13 e u Z gz u Lint = ... + gz Z(uu + d d + cc + s s ) = ... + gz Z(uu + dd + cc + ss) Z 67

(16.469)

(16.473)

(16.474) (16.475)

(16.476)

Aim: To extend the idea of state mixing to neutrinos Learning outcomes: At the end of this lecture you should know about the solar neutrino problem know about atmospheric neutrinos be able to calculate neutrino oscillations due to their mass dierence

17.1

Solar neutrinos

Continuing the theme of state mixing we now examine neutrinos. Ever since their introduction by Pauli, in order to conserve energy, neutrinos hve been rather mysterious, and there are still some rather basic properties of neutrinos that are not understood. Our story here, however, goes back to 1968 with the Homestake experiment of Ray Davis. He and his team put in a herculean eort to measure the number of neutrinos coming from the nuclear fusion events in the Sun, in an eort to test the prediction of Bahcall and collaborators; a more recent version of the predicted ux is given in Fig. 1. There are a number of fusion processes in the Sun that

Figure 1: The predicted ux of solar neutrinos, Bahcall et. al. ApJ, 621, L85 (2005). lead to neutrinos, the largest one being the pp fusion pp de+ e ppe de another key one, from an experimental point of view, is
8

(17.477) (17.478)

B 8 Be e+ e

(17.479)

What we see from Fig. 1 is some rather extreme numbers, with around 60 billion neutrinos going through your small ngernail each second! And yet we do not notice them. The reason we do not feel some huge pressure due to this ux is that they dont interact very strongly with 68

matter. To a very good approximation they al go straight through the Earth without noticing it is there - so a rather delicate experiment is need to measure them. Although most of the solar neutrinos do go through the planet, some of them hit stu, and Davis team relied on the reaction e
37 17 Cl

37 18 Ar

(17.480)

to detect them. They used 615 tons of chlorine (essentially just cleaning fuid), waited for a few weeks, then measured how much of the chlorine had turned into argon. They ended up with 33 argon atoms. This is an astonishing acheivement, and when they reported that this was a factor of three smaller than predicted, nobody was particularly surprised - it would be quite easy to miss a few atoms in 615 tons of cleaning uid. There were, however, some people who took the experiment seriously, one of whom was Pontecorvo who, in 1968, suggested that the neutrinos were all there, but the electron neutrinos had rotated to muon neutrinos. This is just the same basic physics as that behind the kaon decays, and the Cabbibo model - the physical neutrino states are really a superposition of electron, muon and tau neutrinos. And, if they have a dierent mass, then the state oscillates between them. Since the homestake experiment the tanks have got bigger, and the liquids have changed. There are two key experiments that really made people believe that Davis was right, the Solar Neutrino Observatory at Sudbury, using heavy water in its tank which is sensitive to all three neutrinos via e d ppe e,, d npe,, e,, e e,, e (17.481) (17.482) (17.483)

and the superK observatory in Japan, with 50,000 tons of water which is also sensitive to all three neutrino avours through elastic scattering e,, e e,, e (17.484)

17.2

neutrino oscillations

For simplicity we shall shall only consider two neutrino avours, electron and muon, and instead of writing the neutrinos as avour eigenstates, we consider the Hamiltonian eigenstates 1 and 2 , as these are the objects whose evolution is simplest. 1 2 then we know that 1 (t) = 1 (t = 0)eiE1 t , 2 (t) = 2 (t = 0)eiE2 t (17.486) = cos sin sin cos e (17.485)

We know that the solar neutrinos start life as electron neutrinos, as those are what the nuclear fusion events produce, so e (t = 0) = 1, (t = 0) = 0 1 (t = 0) = cos , 2 (t = 0) = sin 69 (17.487) (17.488)

so we nd (t) = sin cos eiE1 t + eiE2 t (E2 E1 )t | (t)|2 = sin2 (2) sin2 2 (17.489) (17.490)

so we see explicitly that the amount of muon neutrino oscillates with time. Now we may make some approximations, namely that because neutrinos are very light compared to their energy we have E2 E1 = p2 + m2 2 p2 + m2 1 (17.491) (17.492) (17.493)

1 1 |p|(1 + m2 /|p|2 ) |p|(1 + m2 /|p|2 ) 2 2 2 1 2 2 m2 m1 2|p| As the neutrinos are highly relavistic we have |p| E, and t x, so P ( ) L = sin2 (2) sin2 2E m2 x , 2L

(17.494) (17.495)

where we should be careful to note that m2 = m2 m2 , not (m2 m1 )2 . 2 1 Thus we nd that for oscillations to occur we need to properties there must be mixing between the neutrino avours, as happens in the quark sector, i.e. = 0. the neutrinos must have dierent masses, m2 = 0 For three neutrinos the basic physics is the same, but the maths gets more complicated, in particular, the 2x2 matrix relating mass eigenstates to avour eigenstates becomes a 3x3 unitary matrix. This matrix is called the MNS (Maki, Nakagawa, Sakata) matrix, and is parametrized by three angles5 cos 13 0 sin 13 cos 12 sin 12 0 1 0 0 0 1 0 sin 12 cos 12 0 UM N S = 0 cos 23 sin 23 0 sin 23 cos 23 sin 13 0 sin 13 0 0 1 The results of all the experiments to date reveal that 12 = 34.5 1.4 , 23 = 43.1 4 , 13 < 10 m2 = (2.4 0.5) 103 eV 2 m2 = (8 0.5) 105 eV 2 , 12 23
5

(17.496) (17.497)

actually theres also a phase, but I shall ignore it.

70

17.3

atmospheric neutrinos

As well as solar neutrinos there is another abundant source of freely-available neutrinos, sourced by cosmic rays. These arise from cosmic rays striking a nucleus of some atom in the atmosphere, producing a shower of (mostly) pions. These pions then decay to muons and muon neutrinos, and these muons then decay to electrons as well as muon and electron neutrino as shown in Fig. 2. The rst, immediately obvious, thing to note from Fig. 2 is that there are twice as many
cosmic ray

pions stuff

e e

Figure 2: The typical shower of particles produced by a cosmic ray. muon neutrinos as electron neutrinos, however, the observation of solar neutrinos by superK shows that roughly equal number of e and are actually observed, with roughly the correct amout of e . Our knowledge of mixing then tells us that the muon neutrinos must be oscillating mostly into tau neutrinos.

18

Lecture eighteen - Higgs mechanism and symmetry breaking

Aim: To discovery a way of giving particles a mass Learning outcomes: At the end of this lecture you should be able to show how symmetry-breaking gives a mass to gauge bosons, and to fermions

18.1

symmetry restoration/breaking

A nice intuitive way to think about symmetry breaking is to think back to ferromagnets, such as a bar magnet, which is a lump of material composed of many atoms, each of which is a tiny magnetic dipole. At high temperatures all these little dipoles point in random directions as the interaction energy between dipoles is overwhelmed by termal energy. This means no direction is picked out, and so we say that the system has a rotational symmetry - when averaged over a macroscopic volume. One way to phrase this is to ask what is the average angle of the dipoles, = , and then for this symmetric state we have = 0. Now suppose we reduce the temperature, such that the thermal energy is below the energy scale of the dipole-dipole interaction. In this situation it is energetically favourable for the little 71

dipoles to point in the same direction and then, as a direction has been selected, we say the rotational symmetry is broken. In terms of the mean angle of the dipoles we thus have = 0 in the symmetry-broken phase. This may not seem relevant to mass generation, but stick with me, I want to highlight the fact that symmetry breaking, which will eventually give us the standard model masses, is in fact a rather common eect.

18.2

superconductivity

Another place where symmetry breaking plays an important role is in superconductors. Here, however, it is not the rotation angle of a magnetic moment thats important, but a rather more abstract quantity - the boson made by pairing two electrons together. In a metal one has lots of electrons oating around and, as they are fermions, they occupy dierent quantum states. If we think of these electrons as being in a lattice, then there can actually be an attractive force between electrons, due to lattice vibrations6 . This allows pairs of electrons to become associated, forming a boson, which can then Bose-condense. We can loosely think of the order parameter in this case as = ee , with a non-zero value in the superconducting phase, and a zero value in the normal phase. Here is where the hint comes that symmetry breaking, = 0, has something to do with mass. It was known for some time that superconductors repel magnetic elds, they cannot penetrate a superconductor, this is why superconductors can hover above magnet, as in Fig. 3. The physical reason for this is that photons aquire an eective mass inside the superconductor,

Figure 3: This shows a superconductor hovering above a magnet. and so it costs them energy to be inside. The idea behind the Higgs mechanism is to give gauge bosons, and other particles, a mass by exploiting a non-zero order parameter, just as photons get an eective mass inside a superconductor when = 0.
6

Superconductivity is not part of this course, so I will not go into detail about the pairing mechanism.

72

18.3

massive elds

Before we give a mass to particles we should gure out what we mean by mass. Going back to the Klein-Gordon Lagrangian we see that the mass of the KG eld comes from the quadratic, non-derivative term of the Lagrangian, 1 1 LKG = m2 2 2 2 For the complex KG eld the mass also came form the quadratic, non-derivative piece LKG,cpx = m2 (18.499) (18.498)

And, nally, we have the Dirac eld, whose mass similarly comes from the quadratic, nonderivative piece of the Dirac Lagrangian L = m (18.500)

So, the theme is that mass is just the quadratic, non-derivative piece of the Lagrangian. At which point we just write down 1 LA = ... m2 A A 2 (18.501)

in order to give the photon a mass. Easy. The problem with doing that is that this term breaks the gauge invariance that we worked so hard to acheive, i.e. if we take A A 1 , then this mass term is not invariant. q However, we have sort of already solved this problem, in the form of covariant derivatives. Recall from lecture 4 that to construct a gauge invariant action, all we need to do is replace derivatives with their covariant counterparts, so the gauge invariant complex KG Lagrangian becomes LKG,cpx = D D m2 Vint = iqA ( ) (q 2 )A A m2 Vint and the important term is this one LKG,cpx = ... (q 2 )A A ... So, if we can arrange for = 0, this leads to an eective mass for the photon of 1 2 m = q 2 . 2 (18.505) (18.504) (18.502) (18.503)

18.4

breaking the symmetry

Before we break the symmetry in the complex KG system, we should make sure we understand what that symmetry is. Recall that the construction of electromagnetism involved a U(1) symmetry, = ei 73 (18.506)

so, for example, if we had = 0 then we would get = , as both vanish. This is what we mean by the symmetry not being broken. On the other hand, if we had = , for some non-zero constant , then clearly = , which is what we mean by the symmetry being broken. So we see that, as with magnetism and superconductivity, the non-vanishing of the order parameter is just a statement about the symmetry being broken. The way we get the symmetry to break is by writing down a suitable interaction potential, the typical choice of the full potential, i.e. quadratic piece plus interaction piece, is V () = (18.507) (||2 2 )2 4 which has a shape somewhat reminiscent of a wine bottle bottom as in Fig. 4. As we can see,

Figure 4: A typical wine bottle, where the dregs of the wine tend to form at the bottom, around the circular minimum. this potential is positive, and is minimized at || = . (18.508) So, the potential is minimized at a particular value for the modulus of , but the angle is not given. This is just like the ferromagnet, the little dipoles in the broken phase could have pointed anywhere, but they must make a choice, and when they do the symmetry is broken. What we have found, then, is a way of creating = 0, which is precisely what we needed in (18.505) to give the photon an eective mass - and we have done it in a way the preserves gauge invariance. 74

18.5

non-Abelian Higgs mechanism

Once the Abelian - U(1) - version is understood it is easy to generalize it to the non-Abelian case, just use the covariant derivatives (9.275), where is now a column vector (9.272). Recall that the Lagrangian has the form L = (D ) D + ... = A A + ... (18.509) (18.510)

and again, for a non-zero value of this term just looks like a mass term for the gauge boson (i.e. quadratic in A ). In the weak sector of the standard model one has to work a little bit harder because the gauge group is a product of groups, SU(2)U(1), but the principle is the same. In fact it is the product structure that allows the W and Z to have a dierent mass, Mw 80GeV , Mz 91GeV , and the photon is not quite the U(1) that appears in SU(2)U(1), but it lives partly in the U(1) and partly in the SU(2). The SU(3) gluons remain massless.

18.6

Yukawa coupling and fermion masses

The Higgs mechanism is not just limited to giving mass to gauge bosons, it also provides a natural way of generating fermion mass. Consider, for simplicity, a model that has a real KG eld and a Dirac eld, and no gauge bosons. We could then couple the KG scalar to the fermion by an interaction term of the form Vint = g. (18.511)

This type of interaction (linear in scalar and quadratic in fermion) is called a Yukawa interaction. Again, we see immediately that if we had = , then m = g. (18.512)

so, once more, if == 0 we generate a mass term for some species - in this case a fermion. This highlights the importance of the Higgs eld. In the standard model the Higgs eld is responsible not just for the masses of the gauge bosons (the W and Z), but also for the fermions; it has a lot of work to do for something we dont actually know exists!

19

Lecture nineteen - the real standard model

Aim: To have a glimpse of what the standard model is actually like Learning outcomes: At the end of this lecture you should be scared

75

19.1

electroweak sector, part I

Over the course of this module we have introduced a number of dierent ideas, all of them with the purpose of understanding the standard model. As you may expect, although the standard model is simple compared to what it could have been, it is still rather complicated. In this lecture we shall bring together some of the concepts from earlier lectures and see how they t into the standard model, or at least the electroweak sector. The rst thing we note is that all the neutrinos that have been observed have been leftchiral, and anti-neutrinos are all right-chiral. We also know that there is a Feynman diagram connecting e e W , but this cant be the full story if the neutrinos are only left-chiral, the e actual Feynman diagram connects e(L) e(L) W . This means that the doublet should e e(L) and Re = e(R) . The fact that the right-hand component really be split up into Le = e(L) of the electron couples only to the photon and Z0 whereas the left-hand component couples to photon, W and Z, means that Le and Re have dierent covariant derivatives. Moreover, it also complicates the possible mass terms as we are not allowed to have Le Le or Re Re because they both vanish - mass terms must contain one of each chirality. However, we are also not allowed to have Re Le as Re contains two fermi elds, but Le contains only one so the coupling does not make sense. Given that the electron does have a mass there must be a resolution to this.

19.2

chiral theories

The term chiral theory refers to theories where the left and right component of the same fermi eld have dierent gauge symmetries. To see how this works lets consider a fermion coupled to a complex KG scalar. The starting point is the ungauged Lagrangian g L = (19.513) 2 which we write as L = L L R R gL R (19.514)

Now that we have it in this form we start thinking about the symmetries. If it werent for the Yukawa term there would be three undependent U(1) symmetries, one each for L , R and . However, for the Yukawa term to be invariant we nd that one of these three is not independent, leading to the full set of symmetry transformations L eiqL L R eiqR R eiq (19.515) (19.516) (19.517)

under which the Lagrangian remains invariant if q = qL qR . So, we are free to give the left and right components a dierent charge when we gauge this symmetry, so long as the scalar has the correct charge. This means that if we promote to (x) then L = L ( iqL A )L R ( iqR A )R ( iq A ) ( iq A ) gL R 76 (19.518)

is invariant. This model then shows us how to make it work in the standard model, even though the left and right fermions transform dierently, we soak up that dierence in the mass term (the Yukawa piece) with a scalar eld that transforms appropriately.

19.3

electroweak sector, part II

We know follow the same proceedure as above, but for the standard model, and start by writing down the ungauged version L = Le Le Re Re g Le Re (19.519) (19.520)

where, at this point, we dont really know what needs to be. There are, again, two symmetries in this lagrangian, but now they are an SU(2) and a U(1) symmetry. The SU(2) comes from Le being a doublet, composed of both electron and neutrino elds; under this symmetry Re remains inert. The correct charge assignments for the standard model are SU (2) : U Le U L e Re Re ei/2 Le ei/2 Le Re ei Re

U (1) :

(19.521)

where U SU (2), so that when we gauge the symmetry we need the following covariant derivatives i (19.522) D = + g B + igA 2 i D Le = Le g B Le + igA Le (19.523) 2 D Re = Re ig B Re (19.524) 1 - this is the Higgs boson 2 that LHC is hoping to nd. Note also that A is the non-Abelian gauge boson associated to i SU(2) - and so is a matrix given by A = Ai - while B is the Abelian gauge boson. 2 and we nd that is a two component complex scalar, =

19.4

electroweak sector, part III

We are still not quite there as we havent come across the photon, the W or the Z0 yet.; to do that we need to break the SU(2)U(1) symmetry. This is achieved, as usual, by giving an expectation value to the scalar eld, conventionally taken as ||2 = 2 /2. So, if we take 0 = the symmetry will be broken, and we nd / 2 2 (D ) D = ... + (gA3 g B )(gA3 g B ) + g 2 (A1 A1 + A2 A2 ) 8 77

and by introducing g = q tan w e = g sin w = g cos w A = cos w B + sin w A3 Z = sin w B + cos w A3 1 W = (A1 iA2 ) 2 we see that this becomes 1 1 + (D ) D = ... m2 Z Z m2 W W m A A z w 2 2 where mA = mW m 1 g2 + g 2 2 1 g = 2 = 0 (19.531) (19.532) (19.533) (19.530) (19.525) (19.526) (19.527) (19.528) (19.529)

i.e. we have just found a theory that gives us a massless photon, A, two W vector bosons of the same mass, and a single heavier Z vector boson. Moreover, we may calculate the fermion covariant derivatives to nd D Re = Re ig cos w A Re + ig sin w Z Re (19.534) 1 W+ + W 2 W+ W i i Le D Le = Le g cos w A sin w Z Le + ig 2 2 2 2 2 3 +ig (19.535) sin w A + cos w Z Le 2 i g 1 sin2 w I + cos2 w 3 Z Le = Le ie (I 3 )A Le + 2 2 cos w ig ig + + ( 1 i 2 )W Le + ( 1 + i 2 )W Le (19.536) 2 2 2 2 and so we nd that, as Re = eR , Re D Re = eR eR ieR eR A + ig sin w eR eR Z e (19.537) (19.538)

showing that the right-handed fermion, Re couples to the photon with strength e, and to the Z-boson with strength g sin w , but not to the W, as required. We also have that, as e(L) Le = , the left-chiral part gives e(L) Le D Le = L L + eL eL ieL eL A + i e g L L Z (19.539) 2 cos w g(2 sin2 w 1) g g + +i eL eL Z + i eL L W + i L eL W 2 cos w 2 2 78

and this one shows that the left-chiral part of the electron couples to the photon with the same strength as the right-chiral part, i.e. e. As well as this we see that the neutrino does not couple to the photon, as should be expected - it has no electric charge. The Z-boson couples to both the electron and the neutrino, and we have a coupling between electron-W-neutrino. What you should take away from this is that although its a bit scary at rst sight, on reection it just contains physics that we have seen before in the module: chirality; Abelian and non-Abelian gauge symmetry; Higgs mechanism. Given that the above Lagrangian is supposed to describe the Universe, I think we have got away with a remarkably simple theory, it could have been a lot worse.

79