62
Part Two · Quantum fields and particles

Interacting fields and the S-matrix

All of particle physics comes down to one question, asked again and again with bigger machines: if these particles go in, what comes out, and how often? The answer is a single object, the S-matrix, short for scattering matrix. Every cross section at the Large Hadron Collider, every decay rate in the Particle Data Group’s tables, is one of its entries. This chapter defines it and shows how to compute it.

On the plate: two particles go in from the left and right; what comes out depends on how strongly they interact.

In and out

Long before a collision, the incoming particles are far apart and move freely: an in state ∣i⟩|i\rangle. Long after, the products are far apart again: an out state ∣f⟩|f\rangle. The S-matrix gives the amplitude for one to become the other, and is unitary, since probabilities must add to one (Chapter 48). Taking out the part where nothing happens and the delta function that enforces energy-momentum conservation leaves the invariant amplitude M\mathcal M:

⟨f∣S∣i⟩=⟨f∣i⟩+i(2π)4δ4 ⁣(∑pf−∑pi) Mfi\langle f|S|i\rangle = \langle f|i\rangle + i(2\pi)^4\delta^4\!\Big(\sum p_f - \sum p_i\Big)\,\mathcal M_{fi}

All the physics is in M\mathcal M. Established

A toy theory

The simplest interacting field theory adds one term to the free scalar Lagrangian of Chapter 58:

L=12(∂ϕ)2−12m2ϕ2−λ4!ϕ4\mathcal L = \tfrac12(\partial\phi)^2 - \tfrac12m^2\phi^2 - \frac{\lambda}{4!}\phi^4

The ϕ4\phi^4 term lets four quanta meet at a point: two in, two out. It is a toy, but the Higgs field has exactly such a self-interaction. Set the coupling λ=\lambda = 1.0 and watch the machine. To lowest order M=−λ\mathcal M = -\lambda, the same for every angle, so the outgoing pair flies off in all directions equally, and the fraction that scatters grows as λ2\lambda^2.

The Dyson series

In the interaction picture of Chapter 45, states change only because of the interaction HI=∫λ4!ϕ4 d3xH_I = \int\frac{\lambda}{4!}\phi^4\,d^3x. The S-matrix is the total evolution from the distant past to the distant future, written with time ordering TT, which puts later operators to the left:

S=Texp⁡ ⁣(−i∫−∞∞HI(t) dt)=1−i ⁣∫ ⁣HI+(−i)22!∬T{HIHI}+⋯\begin{gathered}S = T\exp\!\left(-i\int_{-\infty}^{\infty}H_I(t)\,dt\right)\\ = 1 - i\!\int\! H_I + \frac{(-i)^2}{2!}\iint T\{H_IH_I\} + \cdots\end{gathered}

Each term is one more power of λ\lambda. For weak couplings, the first few terms are enough. The plate shows how the orders build up: the first gives direct scattering; the second, scattering through an intermediate pair. Chapter 63 shows how to evaluate each term, and Chapter 64 how to draw it.

From fields to amplitudes

What can actually be calculated are averages of time-ordered products of fields in the vacuum, the correlation functions ⟨Ω∣T{ϕ(x1)⋯ϕ(xn)}∣Ω⟩\langle\Omega|T\{\phi(x_1)\cdots\phi(x_n)\}|\Omega\rangle. In 1955 Harry Lehmann, Kurt Symanzik and Wolfhart Zimmermann showed how to read M\mathcal M off them: Fourier transform, and pick out the parts that behave like free particles on each external leg. This is the LSZ reduction formula. In practice it says: compute correlation functions with Feynman diagrams, then cut off the external propagators. The details are long; the idea is that scattering amplitudes are what remains when the incoming and outgoing particles are treated as free.

What you now have

In and out states, the S-matrix and the invariant amplitude M\mathcal M, φ⁴ theory with its tree-level amplitude −λ-\lambda, the Dyson series, and the LSZ idea of reading amplitudes from correlation functions.

· · ·

Three questions to carry with you.

Why must S be unitary?tap
The probabilities of all possible outcomes must add to one.
What is the tree-level amplitude in φ⁴ theory?tap
𝓜 = −λ, the same at every angle.
What does the LSZ formula connect?tap
Vacuum correlation functions of fields and S-matrix elements between particle states.

And one to work out. For φ⁴ at tree level, identical particles, the total cross section is σ=λ2/(32πs)\sigma = \lambda^2/(32\pi s). With λ=0.5\lambda = 0.5 and s=1s = 1 (natural units), what is σ\sigma?

σ =

Sources: F. J. Dyson, Phys. Rev. 75, 486 (1949). H. Lehmann, K. Symanzik and W. Zimmermann, Nuovo Cimento 1, 205 (1955). M. E. Peskin and D. V. Schroeder, ch. 4. M. Srednicki, Quantum Field Theory, ch. 5. Checked by code: σ = λ²/(32πs) = 0.002487 for λ = 0.5, s = 1; Dyson coefficients (−i)ⁿ/n!.

Next · Chapter 63
Wick's theorem

Every Feynman diagram is a way of pairing up fields. Wick's theorem counts the pairings.

Turn the page