Here I’ll write the solution of my homeworks in QFT

Problems

1. Classical electromagnetism (with no sources) follows from the action

\[S = \int d^{4}x \Big(-\frac{1}{4}F_{\mu\nu}F^{\mu\nu}\Big),\;\;\;\;\;\text{where}\;F_{\mu\nu}=\partial_{\mu}A_{\nu}-\partial_{\nu}A_{\mu}.\]

(a) Derive Maxwell’s equations as the Euler-Lagrange equations of this action, treating the components $A_{\mu\nu}$ as the dynamical variables. Write the equations in standard form by identifying $E^{i} = -F^{0i}$ and $\epsilon^{ijk}B^{k} = -F^{ij}$.

(b) Construct the energy-momentum tensor for this theory. Note that the usual procedure does not result in a symmetric tensor. To remedy that, we can add to $T^{\mu\nu}$ a term of the form $\partial_{\lambda}K^{\lambda\mu\nu}$, where $K^{\lambda\mu\nu}$ is antisymmetric in its first two indices. Such an object is automatically divergenceless, so

\[\hat{T}^{\mu\nu} = T^{\mu\nu} + \partial_{\lambda} K^{\lambda\mu\nu}\]

is an equally good energy-momentum tensor with the same globally conserved energy and momentum. Show that this construction, with

\[K^{\lambda\mu\nu} = F^{\mu\lambda} A^{\nu},\]

leads to an energy-momentum tensor T that is symmetric and yields the standard formulae for the electromagnetic energy and momentum densities:

\[\varepsilon = \frac{1}{2} (E^2+B^2);\;\;\;\; S = E\times B.\]




2. The complex scalar field. Consider the field theory of a complex-valued scalar field obeying the Klein-Gordon equation. The action of this theory is

\[S = \int d^{4}x (\partial_{\mu}\phi^{\ast}\partial^{\mu}\phi-m^{2}\phi^{\ast}\phi) .\]

It is easiest to analyze this theory by considering $\phi(x)$ and $\phi^{\ast}(x)$, rather than the real and imaginary parts of $\phi(x)$, as the basic dynamical variables.

(a) Find the conjugate momenta to $\phi(x)$ and $\phi^{\ast}(x)$ and the canonical commutation relations. Show that the Hamiltonian is

\[\mathcal{H} = \int d^{3}x (\pi^{\ast}\pi+\nabla\phi^{\ast}\cdot\nabla\phi+m^{2}\phi^{\ast}\phi).\]

Compute the Heisenberg equation of motion for $\phi(x)$ and show that it is indeed the Klein-Gordon equation.

(b) Diagonalize $\mathcal{H}$ by introducing creation and annihilation operators. Show that the theory contains two sets of particles of mass $m$.

(c) Rewrite the conserved charge

\[Q = \int d^{3}x \frac{i}{2} (\phi^{\ast}\pi^{\ast}-\pi\phi)\]

in terms of creation and annihilation operators, and evaluate the charge of the particles of each type.

(d) Consider the case of two complex Klein-Gordon fields with the same mass. Label the fields as $\phi_{a}(x)$, where $a=1,2$. Show that there are now four conserved charges, one given by the generalization of part (c), and the other three given by

\[Q^{i} = \int d^{3}x \frac{i}{2}(\phi^{\ast}_{a}(\sigma^{i})_{ab}\pi^{\ast}_{b}-\pi_{a}(\sigma^{i})_{ab}\phi_{b}),\]

where $\sigma^{i}$ are the Pauli sigma matrices. Show that these three charges have the commutation relations of angular momentum $(SU(2))$. Generalize these results to the case of n identical complex scalar fields.




3. Evaluate the function

\[\bra{0} \phi(x)\phi(y) \ket{0} = D(x-y) = \int \frac{d^3p}{(2\pi)^3}\frac{1}{2E_{P}} e^{-ip(x-y)},\]

for $(x-y)$ spacelike so that $(x-y)^2 = -r^2$, explicitly in terms of Bessel functions.

Here is 30 questions

Introduction to Quantum Field Theory — Oral Exam Study Guide

Course: Introduction to Quantum Field Theory Instructor: Prof. A. D. Polosa Institution: University of Rome “La Sapienza” — A.Y. 2022/23 Source notes: Edited by Salvatore Manfredi D’Angelo


Conventions used throughout

  • Metric (Einstein–Pauli): $\eta_{\mu\nu} = \delta_{\mu\nu} = \mathrm{diag}(1,1,1,1)$, with the 4-th component of any 4-vector imaginary: $x_4 = i x_0 = i c t$. Scalar products then read $x\cdot y = x_\mu y_\mu$ with no sign factors.
  • Natural units $\hbar = c = 1$, except where dimensions are restored explicitly.
  • Feynman slash: $\not{p} = \gamma_\mu p_\mu$.
  • D’Alembertian: $\Box = \partial_\mu \partial_\mu = \nabla^2 - \partial_t^2$ (in this convention $\Box$ for a free massive particle gives $(\Box - m^2)\phi = 0$ for the Klein–Gordon equation, but the notes occasionally write $(\Box + m^2)$ depending on signs of source terms — I’ll keep the convention used in each derivation explicit).

Tier 1 — Warm-up / Foundations

Q1. The Minkowski metric in Einstein–Pauli convention. Why $i$ on the time component?

Answer. In the Einstein–Pauli convention we declare

\[x_\mu = (x_1, x_2, x_3, x_4) = (\vec{x}, i\,c t),\]

so the 4-th component of every 4-vector is purely imaginary. With this trick the metric on Minkowski space is simply the identity:

\[\eta_{\mu\nu} = \delta_{\mu\nu} = \mathrm{diag}(1,1,1,1),\]

and the scalar product of two 4-vectors $x_\mu$ and $y_\mu$ reduces to

\[x \cdot y \;=\; x_\mu y_\mu \;=\; \vec{x}\cdot \vec{y} + (i c t_x)(i c t_y) \;=\; \vec{x}\cdot\vec{y} - c^2 t_x t_y,\]

which is the correct Lorentz invariant.

Why bother? Because every Lorentz scalar is now an ordinary Euclidean dot product, indices can be freely raised and lowered (there is no distinction), and traces of Dirac matrices come out without minus-sign bookkeeping. The cost: any matrix with a single “4” index has imaginary entries (e.g. $\gamma_4$ has $i$’s in it), and we have two $i$’s in the (4,4) component of products — so factors of $i$ migrate from indices into the matrices themselves.

This is the convention Polosa uses throughout, following Veltman’s Diagrammatica.


Q2. Spacelike, timelike, lightlike intervals — and the simultaneity theorem.

Answer. Given two events with separation $\Delta x_\mu$, define $\Delta s^2 = \Delta \vec{x} ^2 - c^2 \Delta t^2$. Then:
Type Condition Physical meaning
Timelike $\Delta s^2 < 0$ A massive particle can connect the two events
Lightlike $\Delta s^2 = 0$ Only a light ray can connect them
Spacelike $\Delta s^2 > 0$ No causal signal can connect them

Simultaneity theorem (spacelike case). Claim: for any two spacelike-separated events, there exists a Lorentz frame in which they are simultaneous.

Proof. In some frame, the events are at $(\vec{x}_1, t_1)$ and $(\vec{x}_2, t_2)$ with $ \Delta \vec{x} ^2 > c^2 \Delta t^2$. Boost along the spatial separation $\Delta \vec{x}$ with velocity $\vec{v}$:
\[\Delta t' = \gamma\!\left(\Delta t - \frac{v\, \Delta x}{c^2}\right).\]

Set $\Delta t’ = 0$ and solve for $v$:

\[v = \frac{c^2 \Delta t}{\Delta x}.\]
The spacelike condition $ \Delta \vec{x} > c \Delta t$ implies $ v < c$, so this velocity is physically attainable. $\blacksquare$

Why it matters in QFT. This theorem is the formal reason that microcausality must be imposed: if two field operators $\phi(x)$ and $\phi(y)$ are spacelike-separated, no frame can put one in the other’s causal past, so they must commute (or anticommute, for fermions) — otherwise some observer would see one influence the other faster than light.

 t
 |     timelike (future)
 |    /
 |   /
 |  / lightcone
 |./_____________ x
 |\
 | \  spacelike  ← simultaneity possible
 |  \
 |   \ timelike (past)

Q3. The principle of least action and the Euler–Lagrange equations.

Answer. Given a Lagrangian $L(q, \dot q)$ describing a system with generalized coordinate $q(t)$, define the action

\[S[q] = \int_{t_1}^{t_2} L(q(t), \dot q(t))\, dt.\]

Principle of least action (Maupertuis). Among all paths $q(t)$ satisfying the boundary conditions $q(t_1) = q_1$, $q(t_2) = q_2$, the physical motion is the one for which $S$ is stationary: $\delta S = 0$.

Derivation of Euler–Lagrange. Vary $q \to q + \delta q$ with $\delta q(t_1) = \delta q(t_2) = 0$:

\[\delta S = \int_{t_1}^{t_2}\!\left(\frac{\partial L}{\partial q}\delta q + \frac{\partial L}{\partial \dot q}\delta \dot q\right) dt \;=\; \int_{t_1}^{t_2}\!\left(\frac{\partial L}{\partial q} - \frac{d}{dt}\frac{\partial L}{\partial \dot q}\right)\delta q\, dt + \underbrace{\left[\frac{\partial L}{\partial \dot q}\delta q\right]_{t_1}^{t_2}}_{=\,0}.\]

The boundary term vanishes by the BCs. Since $\delta q$ is arbitrary in the interior, the integrand must vanish identically:

\[\boxed{\;\frac{d}{dt}\frac{\partial L}{\partial \dot q} - \frac{\partial L}{\partial q} = 0\;}\]

Subtle point Polosa emphasizes. The Euler–Lagrange formulation is given initial conditions ($q(0)$, $\dot q(0)$), whereas the least-action formulation is given boundary conditions ($q(t_1)$, $q(t_2)$). The two are mathematically equivalent but conceptually different: in Maupertuis, “the particle knows where it has to end up” — a non-trivial input which becomes much more natural in path-integral QFT.


Q4. The EM Lagrangian and gauge invariance of $-\tfrac14 F_{\mu\nu}F_{\mu\nu}$.

Answer. The field strength tensor is

\[F_{\mu\nu} = \partial_\mu A_\nu - \partial_\nu A_\mu,\]

antisymmetric in $\mu\nu$. The free EM Lagrangian density is

\[\mathcal{L}_{\text{EM}}^{\text{free}} = -\tfrac{1}{4} F_{\mu\nu} F_{\mu\nu}.\]

Under a gauge transformation $A_\mu \to A_\mu + \partial_\mu S$ (with $S$ an arbitrary scalar function),

\[F'_{\mu\nu} = \partial_\mu(A_\nu + \partial_\nu S) - \partial_\nu(A_\mu + \partial_\mu S) = \partial_\mu A_\nu - \partial_\nu A_\mu + (\partial_\mu \partial_\nu - \partial_\nu\partial_\mu) S = F_{\mu\nu}.\]

The mixed second partial derivatives cancel because they commute, so $F_{\mu\nu}$ is itself gauge invariant — and hence so is $F_{\mu\nu}F_{\mu\nu}$. The $-1/4$ is a normalization choice fixed by demanding that the Euler–Lagrange equations reproduce Maxwell.


Q5. Why $J_\mu A_\mu$ is gauge invariant in the action.

Answer. Adding a coupling to a 4-current,

\[\mathcal{L} = -\tfrac{1}{4} F_{\mu\nu}F_{\mu\nu} + J_\mu A_\mu,\]

the second piece manifestly is not gauge invariant by itself: under $A_\mu \to A_\mu + \partial_\mu S$,

\[J_\mu A_\mu \to J_\mu A_\mu + J_\mu \partial_\mu S.\]

But the physical object is the action $S_{\text{act}} = \int d^4x\, \mathcal{L}$, not the Lagrangian density. Integrate the extra piece by parts:

\[\int d^4 x\; J_\mu \partial_\mu S = \underbrace{\left[J_\mu S\right]_{\text{boundary}}}_{=\,0} - \int d^4x\; (\partial_\mu J_\mu) S.\]

The boundary term vanishes if $S$ falls off at infinity. Therefore the action changes only by $-\int d^4x\, (\partial_\mu J_\mu) S$. If the current is conserved, $\partial_\mu J_\mu = 0$, this vanishes and the action is gauge invariant.

Moral. Gauge invariance of the action requires current conservation. The two are inseparable: the existence of a gauge symmetry forces $\partial_\mu J_\mu = 0$, and a non-conserved current is incompatible with coupling to a gauge field.


Q6. Noether’s theorem in one sentence.

Answer. Every continuous symmetry of the action gives rise to a locally conserved current.

Concrete derivation. Suppose the Lagrangian density $\mathcal{L}(\phi_i, \partial_\mu \phi_i)$ is invariant (up to a total divergence) under an infinitesimal transformation $\phi_i \to \phi_i + \delta \phi_i$. Specifically, $\delta \mathcal{L} = \partial_\mu \Omega_\mu$ for some $\Omega_\mu$. Then by computing $\delta \mathcal{L}$ directly,

\[\delta \mathcal{L} = \frac{\partial \mathcal{L}}{\partial \phi_i}\delta\phi_i + \frac{\partial \mathcal{L}}{\partial(\partial_\mu \phi_i)}\partial_\mu \delta\phi_i.\]

Using the equations of motion $\partial_\mu \tfrac{\partial \mathcal{L}}{\partial(\partial_\mu \phi_i)} = \tfrac{\partial \mathcal{L}}{\partial \phi_i}$, this becomes

\[\delta \mathcal{L} = \partial_\mu\!\left(\frac{\partial \mathcal{L}}{\partial(\partial_\mu \phi_i)} \delta\phi_i\right).\]

Equating to $\partial_\mu \Omega_\mu$ gives the Noether current

\[J_\mu = \frac{\partial \mathcal{L}}{\partial(\partial_\mu \phi_i)}\delta \phi_i - \Omega_\mu, \qquad \partial_\mu J_\mu = 0.\]

Examples in the course:

  • Spacetime translations → energy-momentum tensor $T_{\mu\nu}$.
  • Internal U(1) phase $\phi \to e^{i\alpha}\phi$ → electric current $J_\mu$.
  • Lorentz rotations → angular-momentum tensor $M_{\mu\nu\rho}$.

Tier 2 — Building the Machinery

Q7. From a chain of oscillators to a scalar field.

Answer. Consider $N$ identical masses $m$, joined by springs of constant $K$, on a ring (periodic BCs). Let $q_n(t)$ be the displacement of the $n$-th mass from equilibrium. Newton’s law gives

\[m \ddot q_n = K(q_{n+1} - q_n) - K(q_n - q_{n-1}) = K(q_{n+1} + q_{n-1} - 2 q_n).\]

The Hamiltonian is

\[H = \sum_{n=1}^{N}\!\left[\frac{p_n^2}{2m} + \tfrac{1}{2} K (q_n - q_{n+1})^2\right].\]
   o——K——o——K——o——K——o——K——o   (ring of N masses)
   m     m     m     m     m

Continuum limit. Let lattice spacing be $a$, and define

  • $\phi(x_n, t) \equiv q_n(t)/\sqrt{a}$,
  • $\rho = m/a$ (mass per unit length),
  • $\tau = K a$ (tension, finite as $a \to 0$).

Then $q_{n+1} + q_{n-1} - 2 q_n \to a^2 \partial_x^2 \phi$, and Newton’s equation becomes the wave equation

\[\rho\, \partial_t^2 \phi - \tau\, \partial_x^2 \phi = 0.\]

Add a transverse “restoring” force $-\mu \phi$ at each site (a mass term) and you get

\[(\Box - m^2)\phi = 0,\]

the Klein–Gordon equation for a free scalar field. The original spring constant $K$ becomes the field tension (which controls the speed of propagation), and the added on-site potential becomes the mass.

This derivation is the conceptual key: a field is an infinite collection of coupled oscillators, and the quanta of the field (after second quantization) are particles.


Q8. Inhomogeneous and homogeneous Maxwell equations from the Lagrangian.

Answer. Take

\[\mathcal{L} = -\tfrac14 F_{\mu\nu}F_{\mu\nu} + J_\nu A_\nu,\]

and treat $A_\nu$ as the dynamical variable. Then

\[\frac{\partial \mathcal{L}}{\partial(\partial_\mu A_\nu)} = -F_{\mu\nu}, \qquad \frac{\partial \mathcal{L}}{\partial A_\nu} = J_\nu.\]

The Euler–Lagrange equation $\partial_\mu \tfrac{\partial \mathcal{L}}{\partial(\partial_\mu A_\nu)} - \tfrac{\partial \mathcal{L}}{\partial A_\nu} = 0$ gives

\[\boxed{\;\partial_\mu F_{\mu\nu} = J_\nu\;}\]

— the inhomogeneous Maxwell equations (Gauss + Ampère).

Homogeneous equations. $\vec{\nabla}\cdot \vec{B} = 0$ and Faraday’s law are identities, not equations of motion — they follow automatically from the definition $F_{\mu\nu} = \partial_\mu A_\nu - \partial_\nu A_\mu$. Manipulating $\partial_i (\partial_i A_m - \partial_m A_i) = 0$ and $\varepsilon_{ijk} \partial_t F_{ij} + \ldots = 0$ yields the Bianchi identity

\[\partial_\mu F_{\nu\lambda} + \partial_\nu F_{\lambda\mu} + \partial_\lambda F_{\mu\nu} = 0.\]

To put this in Lagrangian form, define the dual tensor

\[\widetilde{F}_{\mu\nu} = \tfrac{1}{2}\,\varepsilon_{\mu\nu\rho\sigma} F_{\rho\sigma},\]

and the Bianchi identity becomes

\[\partial_\mu \widetilde{F}_{\mu\nu} = 0,\]

which is the “dual” of the inhomogeneous equations with $J_\nu = 0$ — i.e. magnetic monopoles don’t exist.


Q9. The energy-momentum tensor and why it needs symmetrization.

Answer. Applying Noether’s theorem to spacetime translations $x_\mu \to x_\mu + a_\mu$ for any field $\phi_i$ gives the canonical energy-momentum tensor

\[T_{\mu\nu}^{\text{can}} = \frac{\partial \mathcal{L}}{\partial(\partial_\mu \phi_i)}\partial_\nu \phi_i - \delta_{\mu\nu}\mathcal{L}, \qquad \partial_\mu T_{\mu\nu}^{\text{can}} = 0.\]

For the EM Lagrangian $-\tfrac14 F_{\rho\sigma}F_{\rho\sigma}$ this gives

\[T_{\mu\nu}^{\text{can}} = -F_{\mu\rho}\,\partial_\nu A_\rho + \tfrac14 \delta_{\mu\nu} F_{\rho\sigma}F_{\rho\sigma}.\]

Two problems with this:

  1. It is not symmetric in $\mu\nu$.
  2. It is not gauge invariant — it contains $\partial_\nu A_\rho$, not $F_{\nu\rho}$.

For coupling to gravity (Einstein equations want a symmetric tensor) and for sensible physical interpretation (angular momentum density needs $T_{\mu\nu} = T_{\nu\mu}$), one needs to fix this.

Belinfante symmetrization. Add a piece of the form $\partial_\rho \chi_{\rho\mu\nu}$ with $\chi$ antisymmetric in its first two indices — this is divergence-free identically, so doesn’t spoil $\partial_\mu T_{\mu\nu} = 0$. The clever choice is $\chi_{\rho\mu\nu} = F_{\rho\mu}A_\nu$, which gives the symmetric, gauge-invariant tensor

\[\boxed{\;T_{\mu\nu} = F_{\mu\rho}F_{\nu\rho} - \tfrac14 \delta_{\mu\nu} F_{\rho\sigma}F_{\rho\sigma}\;}\]

This is the physically correct stress-energy tensor of electromagnetism, containing the energy density $T_{00} = \tfrac12(E^2 + B^2)$ and the Poynting vector $T_{0i} = (\vec{E}\times\vec{B})_i$.


Q10. The Klein–Gordon equation with a source, and why it motivates propagators.

Answer. Write the scalar Lagrangian with an external source $J(x)$:

\[\mathcal{L} = -\tfrac{1}{2}(\partial_\mu \phi)(\partial_\mu \phi) - \tfrac{1}{2} m^2 \phi^2 + \phi J.\]

The Euler–Lagrange equation gives

\[\boxed{\;(\Box - m^2)\phi(x) = -J(x)\;}\]

This is a linear inhomogeneous PDE, and its general solution is

\[\phi(x) = \phi_0(x) + \int d^4 y\; \Delta(x-y)\,J(y),\]

where $\phi_0$ is any solution of the homogeneous equation and $\Delta(x-y)$ is a Green’s function (a.k.a. propagator) satisfying

\[(\Box - m^2)\Delta(x-y) = -\delta^{(4)}(x-y).\]

In momentum space this is solved instantly:

\[\Delta(x-y) = \int \frac{d^4 k}{(2\pi)^4}\, \frac{e^{i k\cdot(x-y)}}{k^2 + m^2}.\]

Why this is the gateway to QFT. Any interacting theory can be written by promoting the source $J$ to depend on other fields (or on $\phi$ itself), and the resulting non-linear equation can be solved perturbatively — order by order in the coupling, with each order built from convolutions of the free propagator $\Delta$. This is the entire machinery of Feynman diagrams: propagators on internal lines, vertices from the interaction Lagrangian.


Q11. The four contour prescriptions and Feynman’s choice.

Answer. The integral defining $\Delta(x-y)$ involves an integration over $k_0$ (or, in Polosa’s convention, $k_4$) of an integrand with two poles on the real axis at $k_0 = \pm\omega_k$, where $\omega_k = \sqrt{ \vec{k} ^2 + m^2}$. The integral is ambiguous as written; one must specify how to deform the contour around the poles. There are four inequivalent choices:
                   Im k0
                     |
   ─────────●─────────────────●─────────  Re k0
         −ω_k                +ω_k

Going above or below each of the two poles gives $2 \times 2 = 4$ contours:

Choice Behavior Physical name
① Both poles above Retarded propagator
② Both poles below Advanced propagator
③ Above $-\omega_k$, below $+\omega_k$ Feynman propagator
④ Below $-\omega_k$, above $+\omega_k$ Anti-Feynman / Dyson

Equivalently, one shifts the poles by $\pm i\varepsilon$ in the denominator; Feynman’s prescription is

\[\Delta_F(x-y) = \int \frac{d^4 k}{(2\pi)^4}\, \frac{e^{i k\cdot(x-y)}}{k^2 + m^2 - i\varepsilon}.\]

Why Feynman’s choice? Polosa gives two arguments:

  1. Causality/probability. When you compute amplitudes, the propagator appears squared in cross sections. For choices ① and ②, the propagator is identically zero for either $t>0$ or $t<0$ — giving zero or negative probabilities, both unphysical. Choice ③ is the only one that gives a finite, positive probability for both time orderings, naturally interpreted as “forward-propagating particles for $t>0$, and forward-propagating antiparticles for $t<0$ (i.e. backward-propagating particles)”.

  2. Stückelberg–Feynman interpretation. The Feynman propagator is the time-ordered Green’s function

\[\Delta_F(x-y) = \langle 0 | T\{\phi(x)\phi(y)\} | 0 \rangle.\]

This object is exactly what comes out of the S-matrix Dyson expansion (Wick’s theorem), so the choice is imposed by the theory — it isn’t an extra assumption.


Q12. Virtual particles and quantifying virtuality.

Answer. A particle is on-shell if it satisfies the relativistic dispersion relation $p_\mu p_\mu = -m^2$ (in Einstein–Pauli convention; equivalently $E^2 = \vec{p} ^2 + m^2$). A particle is off-shell or virtual if it does not — it appears only on internal propagator lines in a Feynman diagram, never as a free incoming or outgoing state.

Concrete example (elastic e⁻–nucleus scattering). An electron with 4-momentum $K$ exchanges a virtual photon with the nucleus, then comes out with 4-momentum $K’$:

       K           K'
   e⁻ ───→──╳──→─── e⁻
            │
            │ q = K − K'   (virtual photon)
            │
   N  ─────●──────  N

The exchanged 4-momentum is $q_\mu = K_\mu - K’_\mu = (\vec{K}-\vec{K’},\, i(E-E’))$. For elastic scattering, $E = E’$, so

\[q_\mu q_\mu = |\vec{K}-\vec{K'}|^2 > 0.\]
A real photon would have $q^2 = 0$. The virtual photon’s “mass-squared” $q^2 = \vec{K}-\vec{K’} ^2 \cdot \sin^2(\theta/2)$ measures how far off-shell it is. The larger the momentum transfer, the more virtual the photon.

Heisenberg’s uncertainty principle tells us virtual particles can exist only for a time $\Delta t \sim 1/\sqrt{q^2}$ — long enough to mediate the interaction, too short to be detected directly.


Q13. The Dirac $\gamma$ matrices via $\sigma\otimes\sigma$.

Answer. The Dirac algebra is

\[\{\gamma_\mu, \gamma_\nu\} = 2\delta_{\mu\nu} \mathbb{1}.\]

In 4 dimensions the minimal representation is $4\times 4$. A convenient explicit choice — the one Polosa uses — is built from Pauli matrices via Kronecker products:

\[\gamma_i = \sigma_2 \otimes \sigma_i \qquad (i = 1,2,3), \qquad \gamma_4 = \sigma_3 \otimes \mathbb{1}_2.\]

These are all Hermitian: $\gamma_\mu^\dagger = \gamma_\mu$ — a property specific to the Einstein–Pauli convention; in the West-coast/Bjorken–Drell convention only $\gamma^0$ is Hermitian. Explicit forms:

\[\gamma_1 = \sigma_2\otimes\sigma_1 = \begin{pmatrix} 0 & -i\sigma_1 \\ i\sigma_1 & 0 \end{pmatrix}, \quad \gamma_4 = \sigma_3 \otimes \mathbb{1} = \begin{pmatrix} \mathbb{1} & 0 \\ 0 & -\mathbb{1}\end{pmatrix}.\]

Verification of the algebra. Use the Pauli matrix identities ${\sigma_i, \sigma_j} = 2\delta_{ij}\mathbb{1}$ and the property of Kronecker products $(A\otimes B)(C\otimes D) = AC\otimes BD$:

\[\{\gamma_i,\gamma_j\} = \{\sigma_2\otimes\sigma_i,\, \sigma_2\otimes\sigma_j\} = \sigma_2 \sigma_2 \otimes \{\sigma_i,\sigma_j\} = \mathbb{1}\otimes 2\delta_{ij}\mathbb{1} = 2\delta_{ij}\mathbb{1}_4. \checkmark\]

Similarly ${\gamma_i, \gamma_4} = (\sigma_2 \sigma_3 + \sigma_3 \sigma_2)\otimes \sigma_i = 0$, and $\gamma_4^2 = \sigma_3^2\otimes \mathbb{1} = \mathbb{1}$. ✓

Other key properties:

  • $(\gamma_\mu)^2 = \mathbb{1}$ for each $\mu$,
  • Trace of an odd product of $\gamma$’s vanishes,
  • $\mathrm{Tr}(\gamma_\mu \gamma_\nu) = 4\delta_{\mu\nu}$,
  • $\mathrm{Tr}(\gamma_\mu \gamma_\nu \gamma_\rho \gamma_\sigma) = 4(\delta_{\mu\nu}\delta_{\rho\sigma} - \delta_{\mu\rho}\delta_{\nu\sigma} + \delta_{\mu\sigma}\delta_{\nu\rho})$.

These traces are the workhorse of every QED calculation (Appendix 2 of the notes is dedicated to them).


Q14. $\gamma_5$ and chirality.

Answer. Define

\[\gamma_5 = \gamma_1 \gamma_2 \gamma_3 \gamma_4.\]

In the Polosa basis, $\gamma_5 = (\sigma_2\otimes\sigma_1)(\sigma_2\otimes\sigma_2)(\sigma_2\otimes\sigma_3)(\sigma_3\otimes\mathbb{1})$, which after a careful evaluation simplifies (the Polosa notes do this on p.45) to

\[\gamma_5 = -\sigma_1 \otimes \mathbb{1} = \begin{pmatrix} 0 & -\mathbb{1} \\ -\mathbb{1} & 0 \end{pmatrix}.\]

(Conventions differ by signs across textbooks.)

Key properties:

  • ${\gamma_5, \gamma_\mu} = 0$ for $\mu = 1,2,3,4$ — anticommutes with all four $\gamma$’s.
  • $\gamma_5^2 = \mathbb{1}$.
  • $\gamma_5^\dagger = \gamma_5$.

Chirality projectors. Because $\gamma_5^2 = \mathbb{1}$ with eigenvalues $\pm 1$, the operators

\[P_L = \tfrac{1}{2}(\mathbb{1} - \gamma_5), \qquad P_R = \tfrac{1}{2}(\mathbb{1} + \gamma_5)\]

are projectors ($P_L^2 = P_L$, $P_R^2 = P_R$, $P_L P_R = 0$, $P_L + P_R = \mathbb{1}$). They split any Dirac spinor into left-handed and right-handed chiral components:

\[\psi = \psi_L + \psi_R, \qquad \psi_L = P_L \psi,\;\; \psi_R = P_R \psi.\]

Why it matters. The weak interaction couples only to $\psi_L$ for fermions and $\psi_R$ for antifermions — chirality is the language of parity violation. In the ultrarelativistic limit (Q23), chirality coincides with helicity (the projection of spin along momentum), so chirality becomes physically observable.


Tier 3 — Core QFT

Q15. Deriving the Dirac equation and its plane-wave solutions.

Answer.

Strategy. Look for a first-order relativistic wave equation whose solutions automatically satisfy the second-order Klein–Gordon equation. Ansatz:

\[(\gamma_\mu \partial_\mu + m)\psi(x) = 0.\]

Apply $(\gamma_\nu \partial_\nu - m)$ to the left:

\[(\gamma_\nu \partial_\nu - m)(\gamma_\mu \partial_\mu + m)\psi = (\gamma_\nu \gamma_\mu \partial_\nu \partial_\mu - m^2)\psi.\]

Symmetrize over $\mu,\nu$: $\gamma_\nu \gamma_\mu \partial_\nu \partial_\mu = \tfrac12 {\gamma_\nu, \gamma_\mu}\partial_\nu \partial_\mu = \delta_{\mu\nu}\partial_\mu\partial_\nu = \Box$. So

\[(\Box - m^2)\psi = 0. \quad\checkmark\]

Hence the Dirac equation $(\gamma_\mu \partial_\mu + m)\psi = 0$ automatically encodes the relativistic dispersion relation.

Plane-wave solutions. Try $\psi(x) = u(p) e^{i p \cdot x}$ for positive-energy particles:

\[(i\not{p} + m)u(p) = 0.\]

In block form with $\not{p} = \gamma_4 p_4 + \vec\gamma \cdot \vec{p}$ and writing $u = (u_A, u_B)^T$,

\[\begin{pmatrix} -E + m & \vec{\sigma}\cdot\vec{p} \\ -\vec{\sigma}\cdot\vec{p} & E + m \end{pmatrix}\!\begin{pmatrix} u_A \\ u_B\end{pmatrix} = 0.\]

Non-trivial solutions exist if $\det = 0 \Rightarrow E^2 = \vec{p}^2 + m^2$, so $E = \pm\sqrt{\vec{p}^2 + m^2}$. The two positive-energy solutions (labeled $s=1,2$) are

\[u^s(p) = \begin{pmatrix} \chi^s \\ \frac{\vec{\sigma}\cdot \vec{p}}{E+m}\chi^s \end{pmatrix}, \qquad \chi^1 = \begin{pmatrix} 1 \\ 0 \end{pmatrix},\; \chi^2 = \begin{pmatrix} 0 \\ 1 \end{pmatrix}.\]

For negative-energy / antiparticle solutions $\psi(x) = v(p) e^{-i p\cdot x}$:

\[(-i \not{p} + m) v(p) = 0, \qquad v^s(p) = \mathcal{N}_{r,s}\begin{pmatrix} \frac{\vec{\sigma}\cdot\vec{p}}{E+m}\chi^s \\ \chi^s \end{pmatrix},\]

with a sign convention that ensures $v^3 \leftrightarrow u^3$ and $v^4 \leftrightarrow u^4$ are interpreted as antiparticles (Stückelberg–Feynman).

The four solutions together span the spinor space at each momentum.


Q16. Why fermions need anticommutators — the Born–Jordan argument.

Answer. Suppose, naively, you canonically quantize the Dirac field the same way you did the scalar:

\[[a_{\vec k}^s, (a_{\vec k'}^{s'})^\dagger] = \delta_{ss'}\delta_{\vec k\vec k'}.\]

Compute the equal-time commutator $[\psi_a(x), \bar\psi_b(y)]$ at $x_0 = y_0$. The relevant pieces are (using $\sum_s u^s(\vec k)\bar u^s(\vec k) = (-i\not k + m)/2k_0$ and the analogous $v$-spin-sum):

\[[\psi_a(x), \bar\psi_b(y)] = (-\gamma_\mu\partial_\mu + m)_{ab}\, \Delta(x-y) - (-\gamma_\mu\partial_\mu + m)_{ab}\,\Delta(y-x),\]

where $\Delta(x-y)$ is the scalar Pauli-Jordan function. Crucially, the second term comes in with a plus sign, so it does not cancel the first at spacelike separations. We get a non-zero commutator outside the light cone — microcausality is violated.

Born–Jordan fix. Replace the commutator by an anticommutator:

\[\{a_{\vec k}^s, (a_{\vec k'}^{s'})^\dagger\} = \delta_{ss'}\delta_{\vec k\vec k'}, \qquad \{b_{\vec k}^s, (b_{\vec k'}^{s'})^\dagger\} = \delta_{ss'}\delta_{\vec k\vec k'},\]

and then the corresponding field anticommutator becomes

\[\{\psi_a(x), \bar\psi_b(y)\} = (-\gamma_\mu\partial_\mu + m)_{ab}\,\Delta_C(x-y),\]

where $\Delta_C(x-y)$ vanishes for spacelike separations. Microcausality restored. ✓

Connection to Pauli exclusion. Anticommutators automatically imply Pauli exclusion: $(a^\dagger_{\vec k})^2 = \tfrac12{a^\dagger_{\vec k}, a^\dagger_{\vec k}} = 0$, so you cannot create two fermions in the same state. Conversely, if no state can be doubly occupied, then for any state $ K\rangle$ we must have
\[\langle\text{any}|a a^\dagger + a^\dagger a|\text{any}\rangle = 1,\]

which forces ${a, a^\dagger} = 1$. Pauli exclusion ⇔ anticommutation — they are logically equivalent.

This is the spin-statistics theorem in disguise: integer-spin fields commute, half-integer-spin fields anticommute, and the proof rests precisely on demanding microcausality (or, equivalently, positive-norm Hilbert space).


Q17. The spin sum and the numerator of the fermionic propagator.

Answer. Compute

\[\sum_{s=1,2} u^s(p)\,\bar u^s(p) = ?\]

Using the explicit forms of $u^s$ from Q15, with $\bar u = u^\dagger \gamma_4$:

\[u^s \bar u^s = \frac{E+m}{2E}\begin{pmatrix} \chi^s \\ \frac{\vec\sigma\cdot\vec p}{E+m}\chi^s\end{pmatrix}\!\left(\chi^{s\dagger}, \chi^{s\dagger}\frac{\vec\sigma\cdot \vec p}{E+m}\right)\gamma_4.\]

Multiplying out the block matrix and using $\sum_s \chi^s\chi^{s\dagger} = \mathbb{1}_2$,

\[\sum_s u^s \bar u^s = \frac{1}{2E}\begin{pmatrix} (E+m)\mathbb{1} & -\vec\sigma\cdot\vec p \\ \vec\sigma\cdot\vec p & -(E-m)\mathbb{1}\end{pmatrix} = \frac{1}{2E}(-i\not{p} + m).\]

Similarly, $\sum_s v^s \bar v^s = \tfrac{1}{2E}(-i\not p - m)$.

Free fermionic propagator. Solving $(\not\partial + m)\Delta_{pf}(x) = -i\delta^{(4)}(x)\mathbb{1}$ in momentum space and Feynman-prescribing,

\[\boxed{\;\Delta_{pf}(p) = \frac{-i\not{p} + m}{p^2 + m^2 - i\varepsilon}\;}\]

The numerator $(-i\not p + m)$ is exactly the spin-sum $\sum_s u^s \bar u^s$ (up to the $1/(2E)$ factor absorbed into the measure). This is no coincidence: a propagator describes the amplitude for a fermion to travel between two points, and a fermion can carry either spin orientation — the sum over those orientations is the numerator structure.

This is a recurring theme in the course: the polarization sum is the numerator of the propagator, and we’ll see it again for the photon (Q22).


Q18. Interaction picture, Dyson expansion, and the S-matrix.

Answer.

Three pictures of quantum mechanics: | Picture | States | Operators | |————-|————————-|————————| | Schrödinger | Evolve under full $H$ | Time-independent | | Heisenberg | Time-independent | Evolve under full $H$ | | Interaction | Evolve under $H_{\text{int}}$ | Evolve under $H_0$ |

The interaction picture is the natural setting for perturbation theory: split $H = H_0 + H_{\text{int}}$, where $H_0$ is the free (solvable) Hamiltonian and $H_{\text{int}}$ is treated as a small perturbation. Fields in the interaction picture obey the free equations of motion, so they’re easy to work with; the non-trivial dynamics is pushed into the state evolution.

Time-evolution operator. Define $U(t,t_0)$ by $ \psi(t)\rangle_I = U(t,t_0) \psi(t_0)\rangle_I$. It satisfies
\[i\frac{\partial U}{\partial t} = H_{\text{int}}^I(t)\, U(t,t_0), \qquad U(t_0, t_0) = \mathbb{1}.\]

Dyson expansion. Iterating gives

\[U(t,t_0) = \mathbb{1} + (-i)\int_{t_0}^{t}\!dt_1\, H_{\text{int}}^I(t_1) + (-i)^2\!\int_{t_0}^{t}\!dt_1\!\int_{t_0}^{t_1}\!dt_2\, H_{\text{int}}^I(t_1) H_{\text{int}}^I(t_2) + \cdots\]

Restricting the nested integrals to time-ordered regions makes this clumsy; introduce the time-ordering symbol $T{\cdots}$, which orders operators left-to-right by descending time. Then

\[U(t,t_0) = T\,\exp\!\left[-i\int_{t_0}^t H_{\text{int}}^I(t')\, dt'\right] = \sum_{n=0}^\infty \frac{(-i)^n}{n!}\int dt_1\cdots dt_n\, T\{H_{\text{int}}(t_1)\cdots H_{\text{int}}(t_n)\}.\]

S-matrix. Take $t_0 \to -\infty$ and $t \to +\infty$:

\[\boxed{\;S = U(+\infty, -\infty) = T\,\exp\!\left[-i \int_{-\infty}^{+\infty}\!\!d^4 x\;\mathcal{H}_{\text{int}}(x)\right]\;}\]
Matrix elements $\langle f S i\rangle$ between asymptotic free states give scattering amplitudes — the central computable in QFT.

Q19. Tree-level scattering: how Feynman propagators appear in $\langle S\rangle$.

Answer. Take the simplest interacting scalar theory:

\[\mathcal{H}_{\text{int}}(x) = \tfrac{g}{2}\,\phi^2(x)\,\sigma(x),\]

where $\phi$ is one scalar and $\sigma$ another (a “Yukawa-like toy”). Consider $\phi(p_1) + \phi(p_2) \to \phi(p_3) + \phi(p_4)$ via $\sigma$ exchange.

At second order in $g$, the S-matrix element is

\[\langle f | S^{(2)} | i\rangle = \frac{(-i)^2}{2!}\int d^4x\, d^4 y\,\langle f|\,T\{\tfrac g 2 \phi^2(x)\sigma(x)\cdot \tfrac g 2 \phi^2(y)\sigma(y)\}|i\rangle.\]

Wick’s theorem. Time-ordered products of free fields are decomposed into normal-ordered products plus contractions. Each contraction is precisely the Feynman propagator:

\[\langle 0|T\{\sigma(x)\sigma(y)\}|0\rangle = \Delta_F^\sigma(x-y) = \int\!\frac{d^4 k}{(2\pi)^4}\frac{e^{ik\cdot(x-y)}}{k^2 + m_\sigma^2 - i\varepsilon}.\]

The $\phi$ fields contract with the external states $\ket i$ and $\bra f$ via $\langle 0 \phi(x) p\rangle = e^{i p\cdot x}/\sqrt{2E_p V}$. After doing the $d^4 x$ and $d^4 y$ integrals, one gets 4-momentum-conserving deltas at each vertex, and the remaining $d^4 k$ integral pins $k$ to the exchanged momentum.

Result.

\[\mathcal{M} = -i g^2\,\frac{1}{k^2 + m_\sigma^2 - i\varepsilon}, \qquad k = p_1 - p_3.\]
   p1 →─────●─────→ p3
              ╲
               ╲  k = p1 − p3
                ╲ σ-propagator
                 ╲
   p2 →─────●─────→ p4

Origin of the $i\varepsilon$. The $-i\varepsilon$ comes from Feynman’s contour prescription (Q11), which here is not an extra postulate — it is the Wick-contraction definition of the time-ordered product. The S-matrix machinery automatically gives you the Feynman propagator.

Why the $\not{k} + m$ wouldn’t appear here. Because $\sigma$ is a scalar; its propagator has just $1/(k^2 + m^2)$. For exchanged fermions we’d have $(-i \not k + m)/(k^2 + m^2)$; for exchanged photons, $\delta_{\mu\nu}/k^2$.


Q20. Self-energy and the dressed propagator.

Answer. In an interacting theory, the propagator $\Delta_0(p)$ from Q19 is the free (or bare) propagator. Beyond tree level, a propagating particle can emit and reabsorb virtual quanta, modifying its self-energy. Diagrammatically:

   ─●─    +    ─●─⊙─●─    +    ─●─⊙─●─⊙─●─    +   ...
   free       1-loop          2-loop chain

Define the 1PI (one-particle-irreducible) self-energy $\Sigma(p)$ as the sum of all diagrams that cannot be disconnected by cutting a single propagator line. Then the full (dressed) propagator is the geometric series

\[\Delta(p) = \Delta_0(p) + \Delta_0(p)\,\Sigma(p)\,\Delta_0(p) + \Delta_0(p)\,\Sigma(p)\,\Delta_0(p)\,\Sigma(p)\,\Delta_0(p) + \cdots\] \[= \Delta_0(p)\sum_{n=0}^\infty[\Sigma(p)\,\Delta_0(p)]^n = \frac{\Delta_0(p)}{1 - \Sigma(p)\Delta_0(p)}.\]

Substituting $\Delta_0(p) = 1/[i(2\pi)^4(p^2 + m^2)]$:

\[\boxed{\;\Delta(p) = \frac{1}{i(2\pi)^4}\cdot\frac{1}{p^2 + m^2 - \Sigma(p)/i(2\pi)^4}\;}\]

Physical mass. The pole of the full propagator is at $p^2 = -M^2$, where

\[\boxed{\;M^2 = m^2 - \frac{\Sigma(p)}{i(2\pi)^4}\bigg|_{p^2 = -M^2}\;}\]

This $M$ is what we measure in experiments. The parameter $m$ in the Lagrangian (bare mass) is not the physical mass.

Why $\Sigma$ can be complex. If $M^2$ has an imaginary part, $M = M_R + i\Gamma/2$, then the particle is unstable with decay rate $\Gamma$. (This is the spectral interpretation behind resonances like the $\rho$ meson or $Z^0$.)

This is the seed of renormalization. The bare parameters in $\mathcal{L}$ are not observables; only the physical masses, couplings, and field normalizations (after absorbing loop corrections) are.


Q21. Quantizing the EM field in a box.

Answer. In Coulomb gauge ($\vec\nabla\cdot\vec A = 0$, $A_0 = 0$), expand the vector potential in plane waves in a box of volume $V$:

\[\vec A(x) = \sum_{\vec k, \lambda} \frac{1}{\sqrt{2\omega_k V}}\!\left[A_{\vec k\lambda}\,\vec\varepsilon^{(\lambda)}\, e^{i k\cdot x} + A_{\vec k \lambda}^*\,\vec\varepsilon^{(\lambda)*}\, e^{-i k\cdot x}\right],\]

with $\omega_k = \vec{k}$ for photons. The two transverse polarization vectors $\vec\varepsilon^{(\lambda)}$ ($\lambda = 1,2$) satisfy $\vec k\cdot \vec\varepsilon^{(\lambda)} = 0$.

Hamiltonian. Substitute into $H = \tfrac12\int_V (\vec E^2 + \vec B^2)\,d^3x$. Using $\int_V d^3 x\, e^{-i(\vec k - \vec k’)\cdot \vec x} = V\delta_{\vec k\vec k’}$ from the periodic BCs,

\[H = \sum_{\vec k, \lambda} \omega_k\!\left(A_{\vec k\lambda}^* A_{\vec k\lambda} + A_{\vec k\lambda} A_{\vec k\lambda}^*\right) V.\]

Promotion to operators. Identify $A_{\vec k\lambda}/\sqrt{2\omega_k V} \to a_{\vec k\lambda}/\sqrt{2\omega_k V}$ and $A_{\vec k\lambda}^*/\sqrt{2\omega_k V} \to a_{\vec k\lambda}^\dagger/\sqrt{2\omega_k V}$. Demand bosonic commutators $[a_{\vec k\lambda}, a_{\vec k’\lambda’}^\dagger] = \delta_{\vec k\vec k’}\delta_{\lambda\lambda’}$. Then

\[\boxed{\;H = \sum_{\vec k, \lambda} \omega_k\!\left(a_{\vec k\lambda}^\dagger a_{\vec k\lambda} + \tfrac12\right)\;}\]

This is a sum of independent quantum harmonic oscillators, one for each momentum mode and polarization. The $a^\dagger_{\vec k\lambda}$ creates a photon of momentum $\vec k$ and polarization $\lambda$; the ground state $\ket 0$ is annihilated by all $a_{\vec k\lambda}$.

Zero-point energy and the vacuum. The $+\tfrac12$ summed over all modes is the infinite zero-point energy $\sum_{\vec k,\lambda}\omega_k/2$ — formally divergent but, in flat space, irrelevant (only energy differences matter). It is, however, very real in the Casimir effect, where boundary conditions modify the mode sum and produce a measurable force (Milonni’s The Quantum Vacuum is the canonical reference).


Q22. The massive-photon trick and the propagator’s $k_\mu k_\nu / m^2$ term.

Answer.

Setup. A massive vector field $A_\mu$ obeys the Proca equation

\[[\Box \delta_{\mu\nu} - \partial_\mu\partial_\nu - m^2 \delta_{\mu\nu}]\,A_\nu = J_\mu,\]

with three physical polarization states (two transverse + one longitudinal). The propagator is

\[\Delta_{\mu\nu}(k) = \frac{\delta_{\mu\nu} + k_\mu k_\nu / m^2}{k^2 + m^2 - i\varepsilon}.\]

Singularity at $m \to 0$. Naively this diverges. But the resolution is that, for gauge-invariant amplitudes, the $k_\mu k_\nu /m^2$ piece doesn’t contribute.

Why $k_\mu k_\nu /m^2$ drops out — the Ward identity. In any QED process, the photon is coupled to a conserved current $J_\mu$ via $\mathcal{L}{\text{int}} = J\mu A_\mu$. Conservation $\partial_\mu J_\mu = 0$ translates in momentum space to $k_\mu J_\mu(k) = 0$. Therefore, whenever the propagator is sandwiched between external currents (which is the only place it appears in amplitudes), the $k_\mu k_\nu/m^2$ piece is multiplied by $k_\mu J_\mu = 0$ on each side and vanishes.

Conclusion. Take $m \to 0$ safely in the surviving $\delta_{\mu\nu}/k^2$ piece, giving the photon propagator

\[\boxed{\;\Delta_{\mu\nu}^{\gamma}(k) = \frac{\delta_{\mu\nu}}{k^2 - i\varepsilon}\;}\]

This trick is Polosa’s favorite: it lets you study QED as the $m\to 0$ limit of a massive vector field, completely bypassing the subtleties of gauge-fixing (Faddeev–Popov, Gupta–Bleuler etc.) at the level of an introductory course.

Polarization sum for the photon. Analogous to the fermion spin sum, the polarization sum for a massive vector is

\[\sum_{\lambda=1,2,3}\varepsilon_\mu^{(\lambda)}\bar\varepsilon_\nu^{(\lambda)} = \delta_{\mu\nu} + \frac{k_\mu k_\nu}{m^2}.\]

The $k_\mu k_\nu/m^2$ piece is the longitudinal polarization, which decouples in the massless limit thanks to current conservation. For the genuinely massless photon you have only two transverse polarizations.


Tier 4 — Advanced / Synthesis

Q23. Helicity, chirality, and why they coincide ultrarelativistically.

Answer.

Definitions.

  • Helicity $h = \vec\Sigma\cdot\hat p$: projection of spin onto the momentum direction. Eigenvalues $\pm 1/2$ for a spin-½ particle. Helicity is a physical observable.
  • Chirality: eigenvalue of $\gamma_5$. The left-/right-handed components $\psi_{L,R} = P_{L,R}\psi$ are defined irrespective of momentum direction. Chirality is not Lorentz-invariant for massive particles; it is the natural object in the Lagrangian (because $\gamma_5$ commutes with boosts only in the massless limit).
Massive case. Chirality and helicity are different things. A right-chiral spinor in one frame may have left helicity in another frame, because for $ v < c$ one can boost to overtake the particle and reverse its momentum (without changing its spin).
Massless / ultrarelativistic case. For $ \vec p \gg m$, the Dirac equation $(-i\not p + m)u = 0$ reduces (in the chiral basis) to $-i\not p\, u = 0$, which factorizes:
\[\not p = E\gamma^0 - \vec\gamma\cdot \vec p \;\rightarrow\; E(\gamma^0 - \hat p\cdot\vec\gamma).\]

In the Weyl basis, $\gamma_5$ is diagonal with $\pm 1$ blocks, and the Dirac equation forces

\[\gamma_5\, u = h\,u,\]

i.e. chirality eigenvalue = $\pm$ helicity eigenvalue.

                 Particle            Antiparticle
   Right-chiral: right-helicity      left-helicity
   Left-chiral:  left-helicity       right-helicity

Parity violation. The weak interaction is chiral: only $\psi_L$ couples to the $W^\pm$, only $\bar\psi_R$ couples. Because of the chirality-helicity coincidence in the ultrarelativistic limit, this means the W boson sees only left-handed-helicity electrons and right-handed-helicity positrons at high energies — the famous Wu-experiment-style parity violation. Helicity is the signature by which parity violation manifests itself in observable cross sections.


Q24. The Pauli–Lubanski operator and Lorentz-covariant spin.

Answer.

Motivation. In non-relativistic QM, spin is defined as the eigenvalue of $\vec S = \vec\Sigma/2$ projected on some axis, in the rest frame. To get a relativistically covariant definition that reduces to $\vec S$ in the rest frame, we use the Pauli–Lubanski 4-vector:

\[\boxed{\;W_\mu = -\tfrac{1}{2}\varepsilon_{\mu\nu\rho\sigma}\, K_\nu\, \sigma_{\rho\sigma}\;}\]

where:

  • $K_\mu$ is the particle’s 4-momentum,
  • $\sigma_{\rho\sigma} = \tfrac{1}{4}[\gamma_\rho, \gamma_\sigma]$ are the Lorentz generators on spinors,
  • $\varepsilon_{\mu\nu\rho\sigma}$ is the totally antisymmetric Levi–Civita symbol in 4D.

Rest-frame check. In the rest frame, $K_\mu = (0,0,0,im)$. Only $\nu = 4$ contributes:

\[W_\mu = -\tfrac{1}{2}\varepsilon_{\mu 4 \rho \sigma}(im)\sigma_{\rho\sigma} = -\tfrac{im}{2}\varepsilon_{\mu jk}\sigma_{jk} \quad (\text{spatial }\mu = j),\]

and $\varepsilon_{ijk}\sigma_{jk} = 2\sigma_{ij}\hat z_{…} = -i\Sigma_i$ where $\Sigma_i$ is the spin matrix. So

\[W_i = m\,\Sigma_i \,/\,2 \;=\; m\, S_i\, \mathrm{(rest\ frame)}, \qquad W_4 = 0.\]

Thus $W_\mu / m \to (\vec S, 0)$ in the rest frame — the ordinary spin.

In any frame. $W_\mu$ is a genuine Lorentz 4-vector by construction. Its square is a Casimir invariant of the Poincaré group:

\[W_\mu W_\mu = m^2\,s(s+1),\]

independent of frame — this is how spin $s$ is defined Lorentz-invariantly for a massive particle.

Helicity. The helicity operator is

\[h = \frac{W\cdot N}{m} \;\Big|\;N_\mu = (\hat p, 0)\quad\Rightarrow\quad h = \frac{\vec W\cdot \hat p}{m}.\]

In the ultrarelativistic limit $m\to 0$ this becomes singular, but $W_\mu \propto K_\mu$ for massless particles, and helicity is the only invariant.


Q25. The QED cross section $e^- e^+ \to \mu^- \mu^+$.

Answer. This is the textbook QED calculation, and one of the first non-trivial predictions to confront $e^+e^-$ colliders.

Diagrams at tree level.

    e⁻ →─●         ●─→ μ⁻
         ╲         ╱
          ╲ γ*    ╱           "S-diagram" (s-channel)
           ╲    ╱
            ●●●●
           ╱    ╲
          ╱      ╲
    e⁺ →─●        ●─→ μ⁺

Only the s-channel (“S-diagram”) contributes — the t-channel (“T-diagram”) would require $e^-\to\mu^-$ via photon exchange, which violates lepton number. So we have a single diagram.

Amplitude. Using Feynman rules,

\[i\mathcal{M} = \bar v(p_2)(-ie\gamma_\mu) u(p_1)\;\frac{-i\delta_{\mu\nu}}{q^2}\;\bar u(k_1)(-ie\gamma_\nu)v(k_2),\]

with $q = p_1 + p_2$ and $q^2 = -s$ (Mandelstam).

Square and sum over spins. Average over initial spins (factor $1/4$) and sum over final spins. The amplitude squared factorizes into two traces:

\[\overline{|\mathcal{M}|^2} = \frac{e^4}{4 q^4}\,L_{\mu\nu}^{(e)}\,L_{\mu\nu}^{(\mu)},\]

with leptonic tensors

\[L_{\mu\nu}^{(e)} = \mathrm{Tr}[(-i\not p_1 + m_e)\gamma_\mu(-i\not p_2 - m_e)\gamma_\nu],\]

and similarly for the muon.

Carrying out the traces (using the trace identities of Q13):

\[L_{\mu\nu}^{(e)} = 4\left(p_{1\mu}p_{2\nu} + p_{1\nu}p_{2\mu} - \delta_{\mu\nu}(p_1\cdot p_2 + m_e^2)\right).\]

Contracting the two tensors and using $s = -(p_1+p_2)^2 = 2 p_1\cdot p_2$ (for $m_e\to 0$), one obtains (in the high-energy limit $s \gg m_\mu^2$):

\[\overline{|\mathcal{M}|^2} = e^4\,\frac{(1+\cos^2\theta)}{2}\cdot \frac{(4 m^2 + s\,\ldots)}{\ldots},\]

leading after phase-space integration to the total cross section

\[\boxed{\;\sigma(e^+e^-\to\mu^+\mu^-) = \frac{4\pi\alpha^2}{3 s}\;}\]

In practical units (using $\hbar c = 197\,\text{MeV}\cdot\text{fm}$, $1\,\text{barn} = 10^{-24}\text{cm}^2$):

\[\sigma \approx \frac{86.8\,\text{nb}\cdot\text{GeV}^2}{s}.\]

For example at $\sqrt{s} = 10$ GeV, $\sigma \approx 0.87$ nb — a famous benchmark for $e^+e^-$ machines.


Q26. Where QED fails: hints of the electroweak theory.

Answer. The formula $\sigma \propto 1/s$ from Q25 falls monotonically — but experiments at PETRA, LEP, and SLC found that it doesn’t. At $\sqrt s$ near the $Z^0$ mass ($M_Z \approx 91.2$ GeV/$c^2$), the cross section shows a sharp resonance peak about 1000× the QED prediction:

σ (nb)
 │
 │           Z⁰ peak
 │           /\
 │          /  \
 │  QED   /    \    QED
 │  ────╱       \────────
 │
 └────────────────────── √s
         91 GeV

Origin. A second tree-level diagram contributes — exchange of a virtual $Z^0$ boson:

   e⁻ →─●─→─ ─●─→ μ⁻
         ╲   ╱
          ╲ ╱
           Z⁰ (or γ*)
          ╱ ╲
         ╱   ╲
   e⁺ →─●─→─ ─●─→ μ⁺

The $Z^0$ has mass $\sim 91$ GeV — at low $\sqrt s$ its propagator $\sim 1/(s - M_Z^2)$ is heavily suppressed and QED dominates. Near $\sqrt s = M_Z$ the propagator blows up, with the imaginary part of the dressed propagator (Q20) giving a finite width $\Gamma_Z \approx 2.5$ GeV — the famous “$Z^0$ lineshape”.

The coupling. Unlike the pure-vector QED coupling $-i e \gamma_\mu$, the $Z^0$ couples as

\[\frac{i g}{4\cos\theta_W}\gamma_\mu(g_V^f - g_A^f \gamma_5) = \frac{ig}{4\cos\theta_W}\gamma_\mu(\,4\sin^2\theta_W - 1 - \gamma_5\,) \;\text{ for charged leptons}.\]

The $\gamma_5$ piece (axial coupling) is what violates parity — distinguishing the weak interaction from the parity-conserving QED. The Weinberg angle $\theta_W$ parametrizes the mixing between the U(1)$Y$ and SU(2)$_L$ gauge bosons in the unification SU(2)$_L\times$U(1)$_Y \to$ U(1)${\text{em}}$.

Conclusion. QED is the low-energy limit of the electroweak theory: at energies well below $M_Z$, the $Z^0$ propagator can be approximated as $1/M_Z^2$ (a contact term, Fermi’s original four-fermion interaction), and the weak amplitude is suppressed by $s/M_Z^2$ relative to QED. As $\sqrt s \to M_Z$, the suppression goes away and the electroweak unification reveals itself.


Q27. Compton scattering γe⁻ → γe⁻ at tree level.

Answer.

Two diagrams (S and U channels):

   k →    k' →                    k →     k' →
    ~~●~~~~~~              ~~●~~~~~~
       │    │                    │    │
       │    │                    │ (cross)
   p →─●────●─→ p'           p →─●────●─→ p'
      (S-channel)              (U-channel)

Amplitude.

\[i\mathcal{M} = \bar u(p')(-ie\gamma_\nu)\,\frac{-i(\not q_S + m)}{q_S^2 + m^2}(-ie\gamma_\mu)u(p)\,\varepsilon_\mu(k)\varepsilon_\nu^*(k') + (S\leftrightarrow U),\]

with $q_S = p + k$, $q_U = p - k’$.

Sum/average over polarizations. Using $\sum_\lambda \varepsilon_\mu \varepsilon_\nu = \delta_{\mu\nu}$ (Q22) for the photons and the fermion spin sum for the electron:

\[\overline{|\mathcal{M}|^2} = \frac{e^4}{4}\,\frac{1}{(q_S^2+m^2)^2}\,\mathrm{Tr}[\,\ldots\gamma_\mu(\not p + m)\gamma_\nu(\not q_S + m)\gamma_\mu(\not p' + m)\gamma_\nu (\not q_S + m)\ldots\,] + (S\leftrightarrow U) + (\text{interference}).\]

This trace has eight gamma matrices — by hand, it is, in Polosa’s apt phrasing on p.155 of the notes, “simply awful”. One uses identities like $\gamma_\mu \gamma_\alpha\gamma_\mu = -2\gamma_\alpha$ and $\gamma_\mu \gamma_\alpha\gamma_\beta \gamma_\mu = 4\delta_{\alpha\beta}$ to chip it down. The book Diagrammatica by Veltman gives a systematic algorithm.

Modern practice. Use a Computer Algebra System: FORM (the actual Veltman tool, now FORM4 by Vermaseren), or FeynCalc for Mathematica, or xAct. Appendix 2 of the Polosa notes literally provides FORM code.

Result — Klein-Nishina formula. After everything,

\[\boxed{\;\frac{d\sigma}{d\Omega} = \frac{\alpha^2}{2m^2}\left(\frac{\omega'}{\omega}\right)^2\!\left[\frac{\omega}{\omega'} + \frac{\omega'}{\omega} - \sin^2\theta\right]\;}\]

with the kinematic constraint (Compton’s formula)

\[\omega' = \frac{\omega}{1 + (\omega/m)(1-\cos\theta)}.\]

Limits:

  • $\omega \ll m$ (Thomson): $\omega’ \to \omega$, $d\sigma/d\Omega \to (\alpha^2/2m^2)(1 + \cos^2\theta)$ — classical Thomson scattering.
  • $\omega \gg m$: forward peak; total cross section falls as $\sigma \sim (\alpha^2/m\omega)\ln(\omega/m)$.

Q28. Ward identity and gauge invariance of the sum.

Answer. Going back to Compton: replace one of the photon polarization vectors $\varepsilon_\mu(k)$ by $k_\mu$. Gauge invariance demands that the amplitude vanish:

\[k_\mu \mathcal{M}_\mu = 0.\]

This is the Ward identity (more properly, Ward–Takahashi). It is the diagrammatic statement of current conservation.

Diagram-by-diagram check. Consider the S-channel alone:

\[k_\mu \mathcal{M}_\mu^{(S)} = \bar u(p')\gamma_\nu\,\frac{\not q_S + m}{q_S^2 + m^2}\not k\, u(p),\]

with $q_S = p + k$. Use the trick $\not k = (\not q_S - \not p) = (\not q_S + m) - (\not p + m)$ and the Dirac equation $(\not p - im) u(p) = 0$ (incoming electron on shell), $\bar u(p’)(\not p’ - im) = 0$:

\[k_\mu \mathcal{M}_\mu^{(S)} = \bar u(p')\gamma_\nu \!\left[\frac{(\not q_S+m)(\not q_S - \not p)}{q_S^2 + m^2}\right]\!u(p)\]

Some algebra: $(\not q_S+m)\not q_S = q_S^2 + m\not q_S$, and $(\not q_S+m)\not p u(p) = (\not q_S + m)(im) u(p)$ using Dirac. This simplifies to $-\bar u(p’)\gamma_\nu u(p)$ — not zero.

Similarly the U-channel gives $+\bar u(p’)\gamma_\nu u(p)$. The sum vanishes:

\[k_\mu \mathcal{M}_\mu^{(S)} + k_\mu \mathcal{M}_\mu^{(U)} = 0. \quad\checkmark\]

Moral. Individual diagrams violate gauge invariance — only their sum is gauge invariant. This is why one cannot drop “small” diagrams in QED: every diagram of a given order must be included, otherwise the unphysical longitudinal polarization (or the unphysical $k_\mu k_\nu/m^2$ piece of the propagator from Q22) gives non-zero contributions.

This is also the conceptual reason FORM-style tools were invented: gauge-invariance is enforced after summing diagrams, so intermediate expressions are unwieldy and full of cancellations.


Q29. The neutron magnetic moment from virtual particles.

Answer.

The puzzle. The neutron has charge zero, so naively no magnetic moment. Yet experiments measure

\[\mu_p \approx +2.79\,\mu_N, \qquad \mu_n \approx -1.91\,\mu_N,\]

where $\mu_N = e\hbar/(2 m_p c)$. Both proton and neutron have anomalously large magnetic moments compared to the Dirac value, and the neutron’s is negative.

QFT explanation (toy model). Treat the nucleon as a Dirac particle, but allow virtual splittings via virtual pion emission:

    ┌── π⁻ ──┐
    │        │
n ──●────────●── n       ↔  during Δt, neutron is "p + π⁻"
    proton  pp

Total charge is conserved ($n = p + \pi^-$, $0 = +1 + (-1)$). Pion has spin 0, so it carries no magnetic moment. During the time $\Delta t$ when this splitting exists, the magnetic moment is that of a charged proton + π⁻ pair, not of a neutral neutron.

Similarly for the proton: $p \to n + \pi^+$.

Time-averaged moments. Let $\Delta t$ be the fraction of time spent in the split configuration. Then

\[\bar\mu_n = \mu_n^{\text{on-shell}}(1-\Delta t) + (\mu_p + \mu_{\pi^-})\Delta t \approx 0\cdot(1-\Delta t) + (\mu_p - |\mu_{\pi^-}|)\Delta t,\] \[\bar\mu_p = \mu_p^{\text{on-shell}}(1-\Delta t) + (\mu_n + \mu_{\pi^+})\Delta t \approx \mu_p(1-\Delta t) + |\mu_{\pi^+}|\Delta t.\]
Take the difference $ \bar\mu_p - \bar\mu_n $:
\[|\bar\mu_p| - |\bar\mu_n| = \mu_p - 2\mu_p\Delta t + 2|\mu_\pi|\Delta t \approx \mu_p\cdot(1 + \text{small corrections}) \approx 1\,\mu_N.\]

Combined with the observation that $\mu_{\pi^-} < 0$ (since $\mu_\pi$ has same sign as charge), one finds $\bar\mu_n < 0$ — qualitatively matching the experimental signs and magnitudes.

Where this fails. This toy model gets the order of magnitude right but not the precise values. The real story requires the quark model: the proton is $uud$ and the neutron is $udd$. Treating these as constituent quarks (each carrying a Dirac moment $\mu_q = q\hbar/2 m_q$) and using SU(6) spin-flavor wavefunctions gives

\[\mu_p = \tfrac{4}{3}\mu_u - \tfrac{1}{3}\mu_d, \qquad \mu_n = \tfrac{4}{3}\mu_d - \tfrac{1}{3}\mu_u,\]

with the ratio $\mu_p/\mu_n = -3/2 \approx -1.46$ — vs. experimental $-1.46$. Striking agreement from a parameter-free quark model.

Modern QCD lattice calculations now reproduce both moments to $\sim 5\%$.


Q30. Synthesis — Two threads through the course.

The question asks the student to weave together two of the deep recurring ideas through the course. Here is a model answer for Gauge Invariance and Causality as the two threads.


Thread 1: Gauge Invariance — from Lagrangian to observable

Lagrangian level. The free EM Lagrangian $-\tfrac14 F_{\mu\nu}F_{\mu\nu}$ is gauge invariant because $F_{\mu\nu}$ itself is invariant (Q4). Coupling to matter via $J_\mu A_\mu$ remains invariant in the action only if $\partial_\mu J_\mu = 0$ (Q5). Therefore gauge invariance imposes current conservation — which by Noether (Q6) is dual to a continuous internal symmetry of the matter Lagrangian. A gauge symmetry is a local version of a global symmetry, and demanding it dictates the existence of a massless mediator.

Propagator level. Quantizing the photon (Q21) gives only two physical (transverse) polarizations, despite $A_\mu$ being a 4-vector. The “extra” two degrees of freedom (longitudinal + time-like) are removed by gauge fixing. The propagator of a massive vector (Q22) carries a $k_\mu k_\nu/m^2$ piece, which is singular at $m\to 0$ — but the singularity decouples because external currents are conserved ($k_\mu J_\mu = 0$). The massless limit is smooth precisely because of gauge invariance.

Observable level. In any QED process like Compton scattering (Q27–Q28), individual Feynman diagrams violate the Ward identity $k_\mu \mathcal{M}_\mu = 0$. Only their sum is gauge invariant. This is not a calculational nicety — it is required for unitarity (probabilities must be positive and conserved) and for the photon to have only two physical polarizations. Any approximation scheme that drops some diagrams while keeping others will produce gauge-dependent, hence unphysical, cross sections.

Punchline. Gauge invariance is the single thread connecting the symmetry of the action, the structure of the propagator, the cancellation between diagrams, and the positivity of physical cross sections. It is not a postulate one imposes — it is a constraint that must be preserved at every step, because the alternative is a sick theory.


Thread 2: Causality — the iε prescription, microcausality, and the Pauli principle

Propagator level. Causality first appears in choosing the contour for $\Delta_F$ (Q11). Among the four mathematically possible prescriptions, only Feynman’s gives a propagator that vanishes “fast enough” outside the light cone and yields positive transition probabilities. The $i\varepsilon$ in $1/(k^2 + m^2 - i\varepsilon)$ is the operational form of causality at the level of free Green’s functions.

Quantization level. When we quantize fields, we demand that operators at spacelike-separated points cannot influence each other:

\[[\phi(x), \phi(y)] = 0 \quad\text{for }(x-y)^2 > 0 \quad\text{(microcausality).}\]

For scalar fields, the commutator naturally vanishes outside the light cone — bosonic quantization is consistent with causality.

For Dirac fields, commutators fail to vanish for spacelike separations (Q16) — anticommutators are needed. The Born-Jordan argument is purely causal: causality forces antifermion statistics. Therefore the spin-statistics theorem is a consequence of causality: spin-½ ⇒ anticommutation ⇒ Pauli exclusion.

Observable level. Pauli exclusion is responsible for atomic shell structure, electron degeneracy pressure, the existence of neutron stars, the stability of matter. Microcausality also constrains analytic properties of S-matrix elements (dispersion relations, optical theorem) and underwrites the entire formalism of cross sections (Q25, Q27).

Punchline. Causality is not a constraint added on top of QFT — it is built into the theory at every level. The iε in the propagator, the (anti-)commutators of field operators, the Pauli principle for fermions, the dispersion relations for amplitudes, and ultimately the very stability of atoms and stars, are all manifestations of the same demand: information cannot propagate outside the light cone.


Appendix A — Master Diagram of Course Topics

            CLASSICAL MECHANICS         CLASSICAL EM
                  │                          │
                  └──────┬───────────────────┘
                         │
                Lagrangian / Action Principle
                         │
                         ▼
                  NOETHER'S THEOREM
                  (symmetry → current)
                         │
            ┌────────────┼────────────┐
            ▼            ▼            ▼
        Energy-Mom.    Current      Lorentz/
        Tensor T_μν    J_μ          Spin gen.
            │            │            │
            └────────────┼────────────┘
                         │
                         ▼
         FREE FIELD THEORY (oscillator chain → continuum)
                         │
        ┌────────────────┼────────────────┐
        ▼                ▼                ▼
    Scalar φ         Dirac ψ           Vector A_μ
    (Klein-G)        (spin-½)          (photon)
        │                │                │
        ▼                ▼                ▼
  Propagator        Propagator        Propagator
  (Q11, Feynman)    (Q17, fermion)    (Q22, photon)
        │                │                │
        └────────────────┼────────────────┘
                         │
                         ▼
              INTERACTION PICTURE
              (Dyson, S-matrix, Wick)
                         │
                         ▼
              FEYNMAN DIAGRAMS
                         │
        ┌────────────────┼────────────────┐
        ▼                ▼                ▼
   Cross sections    Decay rates    Loop corrections
   (Q25 e⁺e⁻→μ⁺μ⁻)  Γ(A→X)         Self-energy Σ(p)
   (Q27 Compton)                    Renormalization
                                    (Q20)
                         │
                         ▼
             HINTS BEYOND QED
                         │
        ┌────────────────┼────────────────┐
        ▼                ▼                ▼
    Electroweak       Quark model        QCD,
    (Q26 Z⁰ pole)     (Q29 μ_n, μ_p)     Beyond SM

Appendix B — Reading list for each tier

Tier Topics Primary book Sections
1 Lorentz, Lagrangian, EM Jackson, Classical Electrodynamics Ch. 11
1 Action principle, Noether De Wit & Smith Ch. 1–2
2 KG, propagator, Feynman De Wit & Smith Ch. 4
2 Path integral motivation Peskin & Schroeder Ch. 2 + 4
3 Dirac equation, spinors Peskin & Schroeder Ch. 3
3 Quantization of Dirac field Peskin & Schroeder Ch. 3.5
3 EM quantization Milonni, The Quantum Vacuum Ch. 2
4 Feynman rules drill Veltman, Diagrammatica All
4 QED tree processes Peskin & Schroeder Ch. 5
4 Ward identity Peskin & Schroeder §7.4
4 Compton (Klein-Nishina) Peskin & Schroeder §5.5
Extra Italian QM prerequisites Rossetti, Rudimenti di MQ All
Extra Parallel course notes Bonciani, iQFT notes Sapienza ‘22 All

End of study guide. In bocca al lupo for the oral exam!