Preliminaries: What You Need To Know

Quantum eld theory draws heavily from a number of previous courses, both mathematical and physical. Here I summarize the key concepts that you will need throughout this course. This will also serve to set conventions and notations. If youre not comfortable with any of these concepts or terms, you should review a suitable book as soon as possible.

Classical Dynamics
Well need both the Lagrangian and Hamiltonian formulation of dynamical systems. Recall that a physical system is described by coordinates qa , a = 1, . . . , n whose dynamics are governed by a Lagrangian L = L(qa , qa ). The action is dened by
t2

S=
t1

L(qa , qa ) dt

(1.1)

The true path taken by the system is an extremal of the action, i.e. S = 0. This leads to the equations of motion, d dt L qa L =0 qa (1.2)

An important result from classical dynamics that will carry over to eld theories is Noethers theorem. Lets review this theorem: it states that every continuous symmetry gives rise to a conserved quantity. Consider an innitesimal transformation qa = Xa (q, q, t) (1.3)

This is a symmetry if L = 0 for all paths1 . To prove Noethers theorem, consider an arbitrary deformation qa . Then L = where Q= L qa qa (1.5) L L qa + qa qa qa = L d qa dt L qa qa + dQ dt (1.4)

When the equations of motion are obeyed, the term in brackets vanishes and we have L = Q. Since the symmetry transformation qa = Xa (q, q, t) is dened to have L = 0, we learn that when the equations of motion are obeyed, Q = (L/ qa )Xa is a conserved = 0. quantity, i.e. Q
In fact we can relax this condition and ask only that L = dF/dt, so that the action remains unchanged for all paths.
1

To move towards the Hamiltonian formalism, we dene the momentum pa conjugate to qa pa = L qa (1.6)

The Hamiltonian of the system is then dened as H(q, p) = pa qa L (1.7)

where H is to be considered a function of qa and pa rather than qa and qa . The equations of motion are now given by Hamiltons equations qa = H pa , pa = H qa (1.8)

Finally, let us recall the Poisson bracket, an important quantity in classical mechanics which aids the leap to the quantum theory. For any functions f (q, p) and g(q, p) on phase space, the Poisson bracket is dened by {f, g} = f g f g a a qa p p qa (1.9)

b In particular, {qa , qb } = {pa , pb } = 0 while {qa , pb } = a .

Quantum Mechanics
In classical mechanics, the state of a system is determined by a point in phase space, specied by (qa , pa ). In contrast, in quantum mechanics, the state of a system is specied by a vector | in Hilbert space. The path from the classical system to the quantum system proceeds by promoting functions f on phase space to operators f acting on the Hilbert space. The map between the two system is given by the relationship between Poisson brackets and commutators, { , }classical In particular, we have
b [a , qb ] = [a , pb ] = 0 and [a , pb ] = i a q p q

[ , ]quantum

(1.10)

(1.11)

This prescription is known as canonical quantization. Note that a generic classical func tion f (q, p) does not dene a unique quantum operator f due to ordering ambiguities.

 The dynamics of a quantum system is governed by the Hamiltonian operator H. There are two dierent ways of viewing the dynamics. In the Schrdinger picture, the o are time independent. The dynamics states | evolve in time, while the operators O is governed by the Schrdinger equation, o d | S = H | S (1.12) dt where the subscript on | S reminds us that were in the Schrdinger picture. In o contrast, in the Heisenberg picture, the states are time independent, while the operators now change with time. The relationship between the two is given by i | H = eiHt/ | S OH = eiHt/ OS eiHt/ and the operators now evolve by dO H i = [H, OH ] dt (1.15)

(1.13) (1.14)

Note that HS = HH H. The two pictures are entirely equivalent since all correlation functions | O | agree. In quantum eld theory we will jump merrily between these two dierent pictures. We will also employ a third viewpoint which is something of a hybrid of the rst two, known as the interaction picture. In this course our notation will dier slightly from that above. Firstly, we will work in natural units with = 1. Secondly, we will not denote operators with a hat; it should be clear from the context whether were talking about classical objects or quantum objects.

Special Relativity
Minkowski space is parameterized by coordinates x = (x0 , xi ), where = 0, 1, 2, 3 and i = 1, 2, 3. We will often denote x0 as time t, and the spatial three vector xi as x. (On the blackboard, the 3-vector will have a wavy line on the bottom rather than an arrow on top). We will work in units in which the speed of light is set to c = 1. The Minkowski metric is then given by 1 1 = = (1.16) 1 1
so that = . We use ( ) to raise (lower) indices on vectors and tensors.

The relativistic line interval is given by ds2 = dx dx = (dx0 )2 dx dx (1.17)

A relativistic particle of mass m traces out a path x (s) in Minkowski space. The 4-momentum p is dened by p = m dx ds (1.18)

which satises p2 p p = (dx /ds)(dx /ds) = m2 . So writing the 4-momentum p = (E, p) in terms of the energy E and the 3-momentum p, we arrive at the relativistic dispersion relationship for a massive particle, E 2 |p |2 = m2 We will denote the energy of a particle of mass m and 3-momentum p as Ep = |p |2 + m2 (1.20) (1.19)

where we take the +ve square root. This quantity will feature a lot in this course! Another quantity that features prominently is the scalar product p x = p x = Et p x (1.21)

A note on indices: throughout this course, we will employ the summation convention in which repeated indices are summed over. For the spacetime indices it will be crucial to keep track of whether theyre up or down: you should never encounter expressions that look like a b . (For the other indices, such as the a = 1, . . . , n index which appear on the qa , we dont need to be as careful, although its good practice to try!). Also, its worth stressing that repeated indices are dummy indices it doesnt matter what you call them. But its very important that you dont use the same pairs of dummy indices twice. For example, the expression (a b )(c d ) makes no sense even with the brackets! Avoid rampant confusion by writing a b c d to show which pairs are summed over. Mistakenly denoting multiple pairs of dummy indices with the same label will be a very easy trap to fall into in this course.

Fourier Transforms
Well be frequently changing from position space to momentum space using the Fourier transform. As always, we have to decide where the factors of 2 sit. Our convention will be dn k f (x) = f (k) eikx (1.22) (2)n

so that the inverse reads f(k) = dn x f (x) eikx (1.23)

Theres now no excuse for losing track of factors of 2: they always accompany momentum integrals, never position integrals. Remember also that if were doing a Fourier transform over spacetime (as opposed to just space) then k x = k 0 x0 k x.

Dirac Delta Functions

dx (x) = 1

(1.24)

The Fourier transform of the delta-function provides a useful representation that we will make extensive use of:
+

(x) =

dk ikx e 2

(1.25)

In an n-dimensional space, the delta-function is given by (n) (x) = dn k ikx e (2)n (1.26)

Complex Functions
There are a few places in the course where well need to make use of the properties of complex functions. In particular, well need the residue theorem. Let be a positively oriented (i.e. anticlockwise) simple closed contour within (and on) which a function f (z) is analytic except for a nite number of singular points z1 , . . . , zn in the interior of . Then
n

f (z)dz = 2i
i=1

bi

(1.27)

where bi is the residue of f (z) at the singular point zi . The residue of a function with an isolated singular point z0 is dened as the coecient c1 of the Laurent expansion of f (z) about z0 ,

f (z) =
n=0

(z z0 )n +

c2 c1 + + ... z z0 (z z0 )2

(1.28)

for 0 < |z z0 | < R, the radius of convergence.