The Physics Travel Guide

Quantum Field Theory

Prerequisites: Canonical Quantum Field Theory. Full path (20 pages)

Intuitive

Quantum field theory is the framework in which the Standard Model of particle physics and much of modern condensed matter physics are written. Its basic objects are fields: quantities that have a value at every point of space and time, like the electric field or the temperature in a room. Each kind of particle belongs to a field that fills all of space. There is an electron field, a photon field, a quark field for each type of quark, and so on.

A useful picture is a large mattress made of balls connected by springs. If you push one ball, a wave travels across the mattress. In classical physics a wave can have any amplitude. In quantum mechanics the energy of each vibration mode comes in discrete packets, and these packets are what we call particles. A photon is one quantum of vibration of the electromagnetic field; an electron is a quantum of the electron field. This explains why all electrons are exactly identical: they are excitations of the same field.

The theory was developed because quantum mechanics with a fixed number of particles does not fit together with special relativity. At high energies, energy can be converted into mass, so particles are created and destroyed, for example when an electron and a positron annihilate into two photons. A field can gain or lose quanta, so particle creation and annihilation are built into the description from the start. Interactions appear as couplings between fields: one field can lose a quantum while another gains one. Forces between particles, such as the electric repulsion of two electrons, come from the exchange of energy and momentum through such couplings.

In practice physicists use quantum field theory to compute rates of particle collisions and decays, the magnetic moment of the electron, and the behaviour of materials near phase transitions. Its predictions for the electron's magnetic moment agree with measurements to better than one part in a billion.

Concrete

We set ℏ=c=1\hbar = c = 1 and use the metric signature (+,−,−,−)(+,-,-,-). The simplest example is a real scalar field ϕ(x)\phi(x) with Lagrangian density

L=12∂μϕ ∂μϕ−12m2ϕ2.\mathcal{L} = \frac12\partial_\mu\phi\,\partial^\mu\phi - \frac12 m^2\phi^2 .

The Euler-Lagrange equation is the Klein-Gordon equation (∂μ∂μ+m2)ϕ=0(\partial_\mu\partial^\mu + m^2)\phi = 0.

Canonical quantization. The conjugate momentum is π=∂L/∂ϕ˙=ϕ˙\pi = \partial\mathcal{L}/\partial\dot\phi = \dot\phi. We promote ϕ\phi and π\pi to operators with equal-time commutators

[ϕ(t,x⃗),π(t,y⃗)]=i δ3(x⃗−y⃗),[ϕ(t,x⃗),ϕ(t,y⃗)]=[π(t,x⃗),π(t,y⃗)]=0.[\phi(t,\vec x), \pi(t,\vec y)] = i\,\delta^3(\vec x - \vec y), \qquad [\phi(t,\vec x),\phi(t,\vec y)] = [\pi(t,\vec x),\pi(t,\vec y)] = 0 .

The general solution of the field equation is a superposition of plane waves, and the commutators are satisfied by

ϕ(x)=∫d3p(2π)312ωp(ap⃗ e−ip⋅x+ap⃗† eip⋅x),[ap⃗,aq⃗†]=(2π)3δ3(p⃗−q⃗),\phi(x) = \int\frac{d^3p}{(2\pi)^3}\frac{1}{\sqrt{2\omega_p}}\left(a_{\vec p}\,e^{-ip\cdot x} + a_{\vec p}^\dagger\, e^{ip\cdot x}\right), \qquad [a_{\vec p}, a_{\vec q}^\dagger] = (2\pi)^3\delta^3(\vec p - \vec q),

with p0=ωp=p⃗ 2+m2p^0 = \omega_p = \sqrt{\vec p^{\,2} + m^2}. Each Fourier mode is a harmonic oscillator. After normal ordering, which removes an infinite zero-point constant,

H=∫d3p(2π)3 ωp ap⃗†ap⃗,P⃗=∫d3p(2π)3 p⃗ ap⃗†ap⃗.H = \int\frac{d^3p}{(2\pi)^3}\,\omega_p\, a_{\vec p}^\dagger a_{\vec p}, \qquad \vec P = \int\frac{d^3p}{(2\pi)^3}\,\vec p\, a_{\vec p}^\dagger a_{\vec p}.

The vacuum ∣0⟩|0\rangle is defined by ap⃗∣0⟩=0a_{\vec p}|0\rangle = 0. The state ap⃗†∣0⟩a_{\vec p}^\dagger|0\rangle has momentum p⃗\vec p and energy p⃗ 2+m2\sqrt{\vec p^{\,2}+m^2}: it is one particle of mass mm. Applying several creation operators gives multi-particle states, and since the a†a^\dagger commute these states are symmetric under exchange, so the particles are identical bosons. The space spanned by all these states is the Fock space. For spin-12\frac12 fields, such as the electron field described by the Dirac equation, one must use anticommutators instead, and the particles are fermions.

Interactions and perturbation theory. Adding a term such as −λ4!ϕ4-\frac{\lambda}{4!}\phi^4 to L\mathcal{L} makes the theory interacting. It can no longer be solved exactly, and one expands observables in powers of λ\lambda. The objects one computes are correlation functions ⟨Ω∣Tϕ(x1)⋯ϕ(xn)∣Ω⟩\langle\Omega|T\phi(x_1)\cdots\phi(x_n)|\Omega\rangle, from which scattering amplitudes follow by the LSZ reduction formula. Each term of the expansion is represented by a Feynman diagram, built from the propagator

DF(x−y)=⟨0∣Tϕ(x)ϕ(y)∣0⟩=∫d4p(2π)4i e−ip⋅(x−y)p2−m2+iϵD_F(x-y) = \langle 0|T\phi(x)\phi(y)|0\rangle = \int\frac{d^4p}{(2\pi)^4}\frac{i\,e^{-ip\cdot(x-y)}}{p^2 - m^2 + i\epsilon}

and a vertex factor −iλ-i\lambda for each interaction point. Loop diagrams contain divergent momentum integrals, which are handled by renormalization: the parameters in L\mathcal{L} are traded for measured quantities such as the physical mass and coupling.

Path integral. Equivalently, correlation functions are derivatives of the generating functional

Z[J]=∫Dϕ  exp⁡(iS[ϕ]+i∫d4x J(x)ϕ(x)),Z[J] = \int\mathcal{D}\phi\;\exp\left(iS[\phi] + i\int d^4x\,J(x)\phi(x)\right),

which is the starting point for gauge theories, for the Faddeev-Popov treatment of gauge fixing and for lattice simulations after rotating to Euclidean time. The Standard Model is a quantum field theory of this kind, with gauge group SU(3)×SU(2)×U(1)SU(3)\times SU(2)\times U(1), spin-12\frac12 quark and lepton fields, gauge fields and a scalar Higgs doublet.

Abstract

There is no single accepted mathematical definition of an interacting quantum field theory in four dimensions. Several frameworks capture different parts of the structure.

Particles and symmetry. In a relativistic theory the Hilbert space carries a unitary representation of the universal cover of the Poincaré group, R1,3⋊SL(2,C)\mathbb{R}^{1,3}\rtimes SL(2,\mathbb{C}). Wigner's classification identifies one-particle states with irreducible unitary representations of positive energy, labelled by the mass m≥0m \geq 0 and the spin ss for m>0m > 0 or the helicity for m=0m = 0 (leaving aside the massless continuous-spin representations, which do not describe known particles). Given a one-particle space H1\mathcal{H}_1, the free-field Fock space is F(H1)=⨁n≥0SnH1\mathcal{F}(\mathcal{H}_1) = \bigoplus_{n\geq 0} S^n\mathcal{H}_1 for bosons, with the antisymmetric powers ΛnH1\Lambda^n\mathcal{H}_1 for fermions. The free field is an operator-valued distribution on this space built from creation and annihilation operators. For interacting theories Haag's theorem shows that the interacting fields cannot be unitarily related to the free fields on a common Fock space, which is why the interaction picture is only a heuristic.

Axiomatic frameworks. The Wightman axioms characterize a theory by operator-valued tempered distributions on a Hilbert space with a Poincaré invariant vacuum, satisfying covariance, locality and the spectrum condition. The Haag-Kastler framework assigns to each region O\mathcal{O} of spacetime a C*-algebra of observables A(O)\mathfrak{A}(\mathcal{O}), with isotony, Einstein causality and covariance. The Osterwalder-Schrader axioms describe the Euclidean correlation functions; with reflection positivity they reconstruct a Wightman theory. In the Euclidean setting a bosonic theory is ideally a probability measure on the space of distributions S′(Rd)\mathcal{S}'(\mathbb{R}^d). See axiomatic quantum field theory.

Functorial view. Following Atiyah and Segal, a dd-dimensional quantum field theory assigns a vector space Z(Σ)Z(\Sigma) (the space of states) to each closed (d−1)(d-1)-manifold Σ\Sigma and a linear map Z(M):Z(Σin)→Z(Σout)Z(M): Z(\Sigma_{\text{in}})\to Z(\Sigma_{\text{out}}) to each dd-dimensional bordism MM, compatibly with gluing and disjoint union. In other words, it is a symmetric monoidal functor from a bordism category, with whatever geometric structure the theory needs (metric, spin structure, orientation), to vector spaces. For topological field theories no geometric structure beyond the topology is needed, and this definition is fully rigorous; for theories that depend on a metric it serves as a guide.

Perturbative and effective theory. Perturbation theory defines correlation functions as formal power series in the coupling. Gauge theories are quantized perturbatively with BRST or, more generally, Batalin-Vilkovisky methods, where gauge invariance becomes a cohomological condition. Renormalization organizes theories by their behaviour under changes of scale: the space of couplings carries a flow, fixed points of the flow are scale invariant theories, typically conformal field theories, and a quantum field theory with a cutoff is then an effective description valid below that scale. In this view, a continuum QFT is a renormalization group trajectory that starts at an ultraviolet fixed point.

Formulations

The most popular formulations of quantum field theory are


Canonical quantum field theory can be summarized as follows:

Classical Field Theory
canonical quantization
Canonical Quantum Field Theory calculation of transition amplitudes probabilities for scatter processes or decays

The main components of quantum field theory are:

  1. Theory of free fields. Here we neglect any interaction terms and only calculate what different fields do when they are on their own.
  2. Theory of interacting fields. Here we derive the correct terms in our equations that describe interactions using gauge symmetry. Unfortunately, it turns out that we can't solve these equations in closed form but only approximately. There is a powerful mathematical machinery that allows us to calculate results approximately. At the heart of this machinery are the famous Feynman diagrams. In addition, an important trick (the interaction picture) allows us to reuse most of the things we derived for free fields.
  3. Renormalization. The main problem in quantum field theory is that calulcations are technically extremely challenging. One reason for this is that when we calculate things naively we end always end up with the result "infinity". This is physically nonsensical and we need to take greater care. The mathematical machinery that allows us to tame these infinities is known as renormalization.

Recommended Resources

Roadmaps

Serious QFT by Tom Banks

The standard modern textbook is An Introduction to Quantum Field Theory, by Peskin and Schroeder [33]. I recommend especially their wonderful Chapter 5, and all of the calculational sections between 16.5 and 18.5, as well as Chapters 20 and 21. Every serious student of QFT should work out the final project on the Coleman–Weinberg potential, which can be found on page 469. Another standard is Weinberg’s three-volume opus [131]. Here I recommend the marvelous sections on symmetries and anomalies in Volume II. The technical discussions of perturbative effective field theory are invaluable. The section on the Batalin–Vilkovisky treatment of general gauge equivalences is also useful. Volume I should probably be read after completing a first course on the subject. It presents an interesting but idiosyncratic approach to the logical structure of the field. Volume III on supersymmetry is full of gems. In my opinion, it is flawed by an idiosyncratic notation and a tendency to obscure relatively simple ideas in an attempt to give absolutely general discussions. Finally, let me mention a relatively new book by M. Srednicki [168]. I have not gone through it thoroughly, and I do not agree with the author’s ordering of topics, but the pedagogical style of the sections I have read is wonderful. It is clear that everyone in the field will turn to this book for all those nasty little details about minus signs and spinor conventions. I think there is also a chance that it will replace Peskin and Schroeder as a standard textbook.

I have also enjoyed using the books by Bailin and Love [169] and Ramond [170] in my many years of teaching the subject. The books by Itzykson and Zuber [171] and Zinn-Justin [111] are more monographs/encyclopediae than textbooks, but they contain a wealth of detail on specific subjects in field theory that can be found nowhere else.

I mention especially Zinn-Justin’s discussions of the large-order behavior of perturbation theory, of the use of field theory for calculating critical exponents, and of instantons. Other treatments of the field theory/statistical physics interface can be found in the book by Drouffe and Itzykson [172], the marvelous book by Parisi [173], and the book by Ma [174]. Schwinger’s source theory books [175] also belong in the category of non-textbooks, which contain scads of invaluable information about field theory. Among older field-theory books, the second volume of Bjorken and Drell [176] contains lots of useful information, like explicit forms for the spectral representations for higher spin. The books of Nishijima [177] (Chapters 7 and 8), Bogoliubov and Shirkov [178], and even Schweber [179] will reward the really serious student of the subject.

Finally, I want to mention various shorter documents that I think are essential reading for students of quantum field theory. The most important is Kogut and Wilson [126], still the best introduction I know of to Wilson’s profound ideas about renormalization. Next is the 1975 Les Houches lecture-note volume Methods in Field Theory [180], every chapter of which is a gem. Coleman’s book [138] is a collection of lectures on a variety of topics in field theory. I’ve drawn on it heavily for the material about instantons and solitons, but the other lectures are also worth reading.

The contributions of Adler, Weinberg, Zimmermann, and Zumino to the 1970 Brandeis Summer School Lectures, 1 and of Weinberg to the 1964 Brandeis volume, are also worth read- ing [181]. Much of Weinberg’s material reappears in his textbook [131]. Finally, let me mention the reprint volume Selected Papers on Quantum Electrodynamics [182] edited by Schwinger. The contributions of Feynman and Schwinger in particular should be read by every student of field theory.Modern Quantum Field Theory by Banks

How to get started learning QFT as an undergraduate by Flip Tanedo

see https://fliptomato.wordpress.com/2006/12/30/from-griffiths-to-peskin-a-lit-review-for-beginners/

Classical Solutions in Quantum Field Theory
solitons General overview, why should we care about classical solutions?
scalar_1plus1 Kink solution, stability of the kink, boundary conditions, topological charge
Sine-Gordon and Thirring Model Duality
Abelian 2+1 Higgs Model Vortex solution, solitons in gauge theories
Sphalerons
Instantons
qcd_vacuum
Lifecycle of a Quantum Field Theorist
  1. Write down a Lagrangian density LL. This is a polynomial in fields ψ\psi and their derivatives. For example L[ψ]=∂μψ∂μψ−m2ψ2+λψ4L[\psi] = \partial_\mu\psi\partial^\mu\psi - m^2\psi^2 + \lambda\psi^4
  2. Write down the Feynman path integral. Roughly speaking this is ∫ei∫L[ψ]Dψ\int e^{i\int L[\psi]} D\psi The value of this integral can be used to compute "cross sections" for various processes.
  3. Calculate the Feynman path integral by expanding as a formal power series in the "coupling constant" λ\lambda. a0+a1λ+a2λ+⋯a_0 + a_1\lambda + a_2\lambda + \cdots The aia_i are finite sums over Feynman diagrams. Feynman diagrams are a graphical shorthand for finite dimensional integrals.
  4. Work out the integrals and add everything up.
  5. Realise that the finite dimensional integrals do not converge.
  6. Regularise the integrals by introducing a ``cutoff'' ϵ\epsilon (there is usually an infinite dimensional space of possible regularisations). For example ∫R1x2dx⟶∫∣x∣>ϵ1x2dx\int_\mathcal{R} {1\over x^2} dx \longrightarrow \int_{|x|>\epsilon} {1\over x^2} dx
  7. Now we have the series a0(ϵ)+a1(ϵ)λ+⋯a_0(\epsilon) + a_1(\epsilon)\lambda + \cdots Amazing Idea: Make λ\lambda, mm and other parameters of the Lagrangian depend on ϵ\epsilon in such a way that terms of the series are independent of ϵ\epsilon.
  8. Realise that the new sum still diverges even though we have made all the individual aia_i's finite. No good way of fixing this is known. It appears that the resulting series is in some sense an asymptotic expansion.
  9. Ignore step 8, take only the first few terms and compare with experiment.
  10. Depending on the results to step 9: Collect a Nobel prize or return to step 1.

There are many problems that arise in the above steps

  • [Problem 1] The Feynman integral is an integral over an infinite dimensional space and there is no analogue of Lebesgue measure.
  • [Solution] Take what the physicists do to evaluate the integral as its definition.
  • [Problem 2] There are many possible cutoffs. This means the value of the integral depends not only on the Lagrangian but also on the choice of cutoff.
  • [Solution] There is a group GG called the group of finite renormalizations which acts on both Lagrangians and cutoffs.

QFT is unchanged by the action of GG and GG acts transitively on the space of cutoffs. So, we only have to worry about the space of Lagrangians.

  • [Problem 3] The resulting formal power series (even after renormalization) does not converge.
  • [Solution] Work in a formal power series ring.

Lectures on Quantum Field Theory by R. E. Borcherds, A. Barnard

Path to this page

  1. Calculus
  2. Imaginary Numbers
  3. Linear Algebra
  4. Functional
  5. Hilbert Space
  6. Legendre Transformation
  7. Newton's Second Law
  8. Vector Calculus
  9. Newtonian Mechanics
  10. Variational Calculus
  11. Configuration Space
  12. Euler-Lagrange Equations
  13. Special Relativity
  14. Lagrangian Formalism
  15. Phase Space
  16. Classical Field Theory
  17. Hamiltonian Formalism
  18. Schrödinger Equation
  19. Canonical Quantum Mechanics
  20. Canonical Quantum Field Theory