08.10.07 · stat-mech / qft

Wightman axioms (W1–W7)

shipped3 tiersLean: none

Anchor (Master): Wightman, *Phys. Rev.* 101, 860 (1956) (the original Wightman-function paper); Wightman & Gårding, *Ark. Fys.* 28, 129 (1964) (the reconstruction theorem in its definitive form); Streater-Wightman, *PCT, Spin and Statistics, and All That* (1964); Borchers, *Nuovo Cim.* 24, 214 (1962) (the Borchers tensor algebra approach); Haag, *Kgl. Danske Vidensk. Selsk. Mat.-Fys. Medd.* 29, 12 (1955) (the no-go theorem on the interaction picture); Haag & Kastler, *J. Math. Phys.* 5, 848 (1964) (algebraic axioms as the continuous-spectrum alternative); Glimm & Jaffe, *Quantum Physics: A Functional Integral Point of View*, 2nd ed. (Springer, 1987), Ch. 6

Intuition Beginner

A quantum field theory is meant to be an honest list of rules: it should say what the Hilbert space of states is, how relativistic observers see the same physics, what objects play the role of fields, and what locality means in a world where signals cannot travel faster than light. The Wightman axioms are a short list of seven such rules. They aim to be the minimum a relativistic quantum theory must satisfy, independent of any cut-off, lattice, or perturbative recipe.

Why bother with axioms when textbooks already compute scattering amplitudes? Because perturbative calculations rest on conventions whose mutual consistency is not visible inside the calculation. The axioms make those conventions explicit. A model that satisfies the seven rules is automatically Poincaré-covariant, has stable energy, respects microcausality, and supports a vacuum from which every state can be built by acting with fields. A model that fails one rule is telling you which feature of relativistic quantum mechanics it gives up.

An analogy: building codes. Each rule on its own is plain — doors must be a certain width, beams must carry a certain load. The codes together do not say how to design a beautiful house; they rule out houses that fall down. The seven Wightman rules do not say how to write a Lagrangian; they rule out theories that secretly violate relativity, positivity of energy, or causality.

Visual Beginner

A schematic showing four pieces stacked together: a separable Hilbert space drawn as a vertical line; a Poincaré-group rotation acting on it as a unitary; the closed forward light cone in momentum space drawn as a solid cone with the vacuum sitting at its tip; and two spacelike-separated regions on a Minkowski diagram with field operators inside them, joined by an equality sign showing that they commute.

Each piece corresponds to one cluster of axioms. The Hilbert space is W1; the unitary action is W2 and W5; the forward light cone with the vacuum at the tip is W3; the operator-valued distributions are W4; spacelike commutation is W6; and the cyclic property is the statement that polynomials in the field acting on the vacuum sweep out the whole Hilbert space, which is W7.

Worked example Beginner

Check the seven rules on the simplest example, the free real scalar field of mass $m$ in four-dimensional Minkowski space.

Step 1. The Hilbert space is the bosonic Fock space built from the one-particle space of square-integrable functions on the mass shell. This is a separable Hilbert space, so W1 holds.

Step 2. The Poincaré group acts on each one-particle state by the standard mass- $m$ , spin-zero representation, and on the full Fock space by taking symmetric tensor powers and pasting them together. The result is a continuous unitary action of the Poincaré group, so W2 holds.

Step 3. The energy and momentum operators are read off from the one-particle representation. Their joint spectrum is the closed forward light cone, with the vacuum (the zero-particle state) sitting at the tip. The vacuum is the only Poincaré-invariant unit vector up to a phase, so W3 holds.

Step 4. The free field is built from creation and annihilation operators integrated against test functions. The pairing gives an operator-valued tempered distribution acting on a dense subspace of Fock space, so W4 holds.

Step 5. Translating the test function translates the operator in exactly the way required, and a Lorentz transformation on the test function matches the corresponding unitary on the field. So W5 holds.

Step 6. A direct computation of the commutator of two free fields evaluates to a multiple of the Pauli-Jordan distribution, which vanishes outside the light cone. So spacelike-separated test functions give commuting operators, and W6 holds.

Step 7. Polynomials in creation operators acting on the vacuum produce every n-particle state, so the linear span of all such vectors is dense in Fock space. W7 holds.

What this tells us: the free scalar field is a complete worked example of a Wightman quantum field theory in four-dimensional Minkowski space. The whole list of axioms is satisfied at once, and the construction is the model every interacting candidate is measured against.

Check your understanding Beginner

Formal definition Intermediate+

Let $P_{+}^{↑} = R^{1, 3} ⋊ L_{+}^{↑}$ be the proper orthochronous Poincaré group, and let $P = R^{1, 3} ⋊ SL (2, C)$ be its universal cover, with $SL (2, C)$ the double cover of the restricted Lorentz group $L_{+}^{↑} = SO^{+} (1, 3)$ . Write $S (R^{1, 3})$ for the Schwartz space of rapidly decreasing test functions, with the topology of the seminorm family $∥ f ∥_{α, β} = sup_{x} ∣ x^{α} (D^{β} f) (x) ∣$ .

A Wightman quantum field theory consists of the following data, subject to seven conditions:

(W1) Hilbert space. A separable complex Hilbert space $H$ .

(W2) Poincaré covariance. A strongly continuous unitary representation $U : P \to U (H)$ , written $U (Λ, a)$ for $Λ \in SL (2, C)$ and $a \in R^{1, 3}$ .

(W3) Spectrum condition and vacuum. The infinitesimal generators of translations $P^{μ} = - i \partial_{a^{μ}} U (1, a) ∣_{a = 0}$ are commuting self-adjoint operators whose joint spectrum is contained in the closed forward light cone $\overline{V_{+}} = {p \in R^{1, 3} : p^{0} \geq 0, p \cdot p \geq 0}$ . There is a unit vector $∣Ω ⟩ \in H$ , the vacuum, such that $U (Λ, a) ∣Ω ⟩ = ∣Ω ⟩$ for every $(Λ, a)$ , and $∣Ω ⟩$ is unique up to a phase.

(W4) Fields. A finite collection of fields $ϕ_{j}$ , $j = 1, \dots, N$ , each carrying a finite-dimensional representation of $SL (2, C)$ on indices. For every $f \in S (R^{1, 3})$ , $ϕ_{j} (f)$ is an unbounded operator on $H$ defined on a dense common invariant domain $D \subset H$ that contains $∣Ω ⟩$ . The map $f \mapsto ϕ_{j} (f) ψ$ is continuous from $S (R^{1, 3})$ to $H$ for every $ψ \in D$ , so each $ϕ_{j}$ is an operator-valued tempered distribution.

(W5) Field-translation covariance. For $(Λ, a) \in P$ and $f \in S (R^{1, 3})$ , $$ U(\Lambda, a), \phi_j(f), U(\Lambda, a)^{-1} = \sum_k S_{jk}(\Lambda^{-1}), \phi_k(f_{\Lambda, a}), $$ where $f_{Λ, a} (x) = f (Λ^{- 1} (x - a))$ and $S (Λ)$ is the chosen finite-dimensional representation of $SL (2, C)$ .

(W6) Locality (microcausality). Whenever the supports of $f, g \in S (R^{1, 3})$ are spacelike separated, $$ [\phi_j(f), \phi_k(g)]_\pm = 0 $$ on $D$ , with a commutator for bosonic fields and anticommutator for fermionic ones, the choice determined by the spin of the field as fixed by W4.

(W7) Cyclicity of the vacuum. The linear span of the vectors $ϕ_{j_{1}} (f_{1}) \dots ϕ_{j_{n}} (f_{n}) ∣Ω ⟩$ , ranging over all finite $n$ and all choices of indices and test functions, is dense in $H$ .

The vacuum expectation values $$ W_n^{j_1 \cdots j_n}(x_1, \ldots, x_n) = \langle \Omega | \phi_{j_1}(x_1) \cdots \phi_{j_n}(x_n) | \Omega \rangle $$ are tempered distributions on $S (R^{4 n})$ , called the Wightman functions of the theory. They are continuous multilinear functionals of the $n$ test functions, Lorentz-covariant, supported on the closed forward light cone in each spectral variable, locality-symmetric, positive in the Borchers-algebra sense, and they satisfy cluster decomposition at spacelike infinity in the connected vacuum.

Counterexamples to common slips

The fields are operator-valued distributions, not operators at a point. The expression $ϕ (x)$ is a heuristic shorthand for the distributional kernel of $ϕ (f) = \int ϕ (x) f (x) d^{4} x$ . Treating $ϕ (x)$ as an honest operator is a category error: the would-be operator $ϕ (x)$ has infinite norm even on the vacuum, since $⟨ Ω∣ ϕ (x) ϕ (x) ∣Ω ⟩ = W_{2} (x, x)$ is the diagonal value of a distribution and is generally divergent.
Uniqueness of the vacuum is part of W3, not a separate axiom. A model with multiple Poincaré-invariant unit vectors is not a Wightman theory in the strict sense; it is a direct sum of superselection sectors, each of which is.
W6 commands a commutator at spacelike separation, not equal times. Equal-time canonical commutation relations are a much stronger assumption that the free field happens to satisfy but that interacting Wightman theories need not. Microcausality is the relativistic content of locality.

Key theorem with proof Intermediate+

Theorem (Wightman-Gårding reconstruction; Wightman 1956 ^{[Wightman 1956]}, Wightman-Gårding 1964 ^{[Wightman-Gårding 1964]}). Let ${W_{n}}_{n \geq 0}$ be a sequence of tempered distributions $W_{n} \in S^{'} (R^{4 n})$ satisfying

(i) Hermiticity: $\overline{W_{n} (x_{1}, \dots, x_{n})} = W_{n} (x_{n}, \dots, x_{1})$ ;

(ii) Poincaré invariance: $W_{n} (Λ x_{1} + a, \dots, Λ x_{n} + a) = W_{n} (x_{1}, \dots, x_{n})$ for all $(Λ, a) \in P$ ;

(iii) Spectrum condition: the Fourier transform of $W_{n}$ in the difference variables $ξ_{k} = x_{k} - x_{k + 1}$ is supported in the closed forward light cone $\overline{V_{+}}^{n - 1}$ ;

(iv) Locality: $W_{n} (\dots, x_{j}, x_{j + 1}, \dots) = W_{n} (\dots, x_{j + 1}, x_{j}, \dots)$ when $(x_{j} - x_{j + 1})$ is spacelike;

(v) Positivity: for every finite sequence of test functions $f_{0} \in C$ , $f_{1} \in S (R^{4}), \dots, f_{N} \in S (R^{4 N})$ , $$ \sum_{m, n = 0}^{N} \int W_{m+n}(\bar{y}_m, \ldots, \bar{y}_1, x_1, \ldots, x_n), \overline{f_m(y_1, \ldots, y_m)}, f_n(x_1, \ldots, x_n), dy, dx \geq 0; $$

(vi) Cluster decomposition: $W_{m + n} (x_{1}, \dots, x_{m}, x_{m + 1} + a, \dots, x_{m + n} + a) \to W_{m} (x_{1}, \dots, x_{m}) W_{n} (x_{m + 1}, \dots, x_{m + n})$ in $S^{'}$ as $a$ tends to spacelike infinity.

Then there exists a Wightman quantum field theory $(H, U, ∣Ω ⟩, {ϕ_{j}})$ whose Wightman functions are the given ${W_{n}}$ , and this realisation is unique up to unitary equivalence.

Proof. The construction is the Gelfand-Naimark-Segal (GNS) construction applied to the Borchers tensor algebra. Let $B = ⨁_{n \geq 0} S (R^{4 n})$ be the Borchers algebra, where the product on tensor symbols is $$ (f \otimes g)(x_1, \ldots, x_{m+n}) = f(x_1, \ldots, x_m), g(x_{m+1}, \ldots, x_{m+n}), $$ and the conjugation is $f^{*} (x_{1}, \dots, x_{n}) = \overline{f (x_{n}, \dots, x_{1})}$ . The sequence ${W_{n}}$ packages as a continuous linear functional $$ \omega : \mathcal{B} \to \mathbb{C}, \qquad \omega(f) = \sum_n \int W_n(x_1, \ldots, x_n), f_n(x_1, \ldots, x_n), dx, $$ with $ω (1) = W_{0} = 1$ .

Step 1: GNS Hilbert space. Hermiticity (i) makes $ω$ a state on the $*$ -algebra $B$ . Positivity (v) is the statement that $⟨ f, g ⟩_{ω} := ω (f^{*} g)$ is a positive semidefinite Hermitian form on $B$ . Quotient $B$ by its null subspace $N = {f \in B : ω (f^{*} f) = 0}$ , then complete in the induced inner product to obtain a Hilbert space $H$ . The class of $1 \in B$ gives a distinguished unit vector $∣Ω ⟩ \in H$ with $⟨ Ω∣Ω ⟩ = ω (1) = 1$ . Separability of $H$ follows from separability of $S (R^{4 n})$ in its Fréchet topology, which gives W1.

Step 2: Field operators. For $f \in S (R^{4})$ , define $ϕ (f)$ on the dense domain $D = B / N \subset H$ by $ϕ (f) [g] = [f \otimes g]$ . The product structure of $B$ ensures that $ϕ (f)$ maps $D$ into itself. Continuity in $f$ in $S (R^{4})$ of the map $f \mapsto ϕ (f) ψ$ for $ψ \in D$ is a consequence of the temperedness of the $W_{n}$ , since the matrix elements $⟨[h] ∣ ϕ (f) [g]⟩ = ω (h^{*} (f \otimes g))$ are continuous multilinear functionals of $f, g, h$ . So $ϕ$ is an operator-valued tempered distribution, giving W4.

Step 3: Poincaré representation. For $(Λ, a) \in P$ , define $$ U(\Lambda, a)[f]_n = [\Lambda \cdot f]_n, \qquad (\Lambda \cdot f)_n(x_1, \ldots, x_n) = f_n(\Lambda^{-1}(x_1 - a), \ldots, \Lambda^{-1}(x_n - a)). $$ Poincaré invariance (ii) of the $W_{n}$ makes $U (Λ, a)$ an isometry on $B / N$ , hence extends uniquely to a unitary on $H$ . The Poincaré group acting continuously on $S (R^{4 n})$ gives strong continuity of $U$ , and the vacuum $∣Ω ⟩$ is fixed because $Λ \cdot 1 = 1$ . This gives W2, and the field-translation covariance W5 is the same statement on the level of $ϕ (f)$ .

Step 4: Spectrum condition. Stone's theorem applied to the continuous one-parameter unitary group $a \mapsto U (1, a)$ produces the self-adjoint translation generators $P^{μ}$ . The spectrum condition (iii) on the $W_{n}$ translates by Fourier transform into the statement that the joint spectral measure of $(P^{0}, \dots, P^{3})$ is supported in $\overline{V_{+}}$ . Uniqueness of the vacuum follows from cluster decomposition (vi): an invariant vector different from $∣Ω ⟩$ would force the $W_{n}$ to fail to cluster, contradicting (vi). This gives W3.

Step 5: Locality. The symmetry property (iv) of the $W_{n}$ at spacelike separation transfers to the commutator $[ϕ (f), ϕ (g)]$ vanishing on $D$ whenever $supp f$ and $supp g$ are spacelike separated. This gives W6 in the bosonic case; the fermionic case uses antisymmetric W6 and the corresponding sign in (iv).

Step 6: Cyclicity. The dense domain $D = B / N$ is by construction the linear span of $ϕ (f_{1}) \dots ϕ (f_{n}) ∣Ω ⟩$ , so cyclicity W7 holds by definition of $H$ .

Step 7: Uniqueness. Any two Wightman realisations $(H_{i}, U_{i}, ∣ Ω_{i} ⟩, ϕ_{i})$ with the same Wightman functions admit a unitary intertwiner $V : H_{1} \to H_{2}$ defined on the cyclic domain by $V (ϕ_{1} (f_{1}) \dots ϕ_{1} (f_{n}) ∣ Ω_{1} ⟩) = ϕ_{2} (f_{1}) \dots ϕ_{2} (f_{n}) ∣ Ω_{2} ⟩$ . Equality of matrix elements via the $W_{n}$ ensures $V$ is well-defined and isometric; cyclicity ensures $V$ extends to a unitary; field-translation covariance ensures $V$ intertwines $U_{1}$ and $U_{2}$ . So the realisation is unique up to unitary equivalence. $□$

Bridge. The reconstruction theorem builds toward every constructive quantum field theory program by reducing the entire content of a relativistic quantum theory to a sequence of tempered distributions and six algebraic conditions on them. The central insight is exactly the Gelfand-Naimark-Segal mechanism transplanted from $C^{*}$ -algebras to the Borchers tensor algebra, with positivity playing the role of the GNS positive functional and the Poincaré action lifting from the algebra to the Hilbert-space completion. This is exactly the foundational reason axiomatic QFT is a meaningful subject: a quantum field theory is the same datum as a sequence of vacuum expectation values, and any procedure that produces such a sequence with the six properties produces a quantum field theory. The bridge is the recognition that this reconstruction generalises to the Euclidean side as the Osterwalder-Schrader theorem, which appears again in 08.10.05 as the analytic-continuation structure of the Feynman propagator and identifies Wightman functions with Schwinger functions in the spacelike region. Putting these together, the Wightman framework gives a relativistic theory in Minkowski signature; Osterwalder-Schrader gives a probabilistic theory in Euclidean signature; and the two are dual to one another, with the Wightman reconstruction theorem and the Osterwalder-Schrader reconstruction theorem as the two halves of the same bridge. The foundational reason this works is that positivity, Poincaré invariance, the spectrum condition, locality, and cluster decomposition are exactly the relativistic-invariant translations of reflection positivity, Euclidean invariance, and clustering on the Euclidean side, with the central insight being that one analytic continuation identifies the two.

Exercises Intermediate+

Exercise 1 (easy, numeric).

The two-point Wightman function $W_{2} (x, y) = ⟨ Ω∣ ϕ (x) ϕ (y) ∣Ω ⟩$ of a Wightman quantum field theory is a tempered distribution on $R^{4} \times R^{4}$ . By translation invariance, it depends on $(x - y)$ alone. Its Fourier transform in $ξ = x - y$ is a tempered distribution $\hat{W}_{2} (p)$ on $R^{4}$ . What is the support of $\hat{W}_{2} (p)$ for a Wightman theory?

Hint

Apply the spectrum condition W3. The Fourier transform of $W_{2}$ in the difference variable encodes the spectral resolution of the energy-momentum operator inserted between the two field operators.

Answer

The closed forward light cone $\overline{V_{+}} = {p : p^{0} \geq 0, p \cdot p \geq 0}$ . Inserting a complete set of energy-momentum eigenstates $\int d μ (p) ∣ p ⟩ ⟨ p ∣$ between the two fields gives $\hat{W}_{2} (p) = \int d ρ (m^{2}) θ (p^{0}) δ (p^{2} - m^{2})$ via the Källén-Lehmann representation, with $ρ \geq 0$ a positive measure on $[0, \infty)$ called the spectral function. Support in $\overline{V_{+}}$ is the spectrum condition W3 translated by Fourier transform. Rubric: full credit for $\overline{V_{+}}$ .

Exercise 2 (easy, multiple choice).

Which axiom of the Wightman framework is the precise relativistic formulation of locality?

A. W2 (Poincaré covariance)
B. W3 (spectrum condition)
C. W6 (microcausality)
D. W7 (cyclicity of the vacuum)

Hint

Locality in a relativistic theory means that observables in regions that cannot causally influence one another should commute.

Answer

C. W6 (microcausality). W6 states that field operators smeared against test functions with spacelike-separated supports commute (or anticommute for fermions). This is the precise relativistic formulation of locality: regions that lie outside one another's light cones cannot causally influence one another, so operators localised there must be compatible. The equal-time canonical commutation relations of Lagrangian QFT are a stronger condition that the free field happens to satisfy but interacting Wightman theories need not. Rubric: full credit for C.

Exercise 3 (medium, symbolic).

Compute the two-point Wightman function of the free real scalar field of mass $m$ in four-dimensional Minkowski space, and verify that it satisfies the spectrum condition.

Hint

The free scalar field expands as a Fourier integral over the positive mass shell. The vacuum expectation $⟨ Ω∣ ϕ (x) ϕ (y) ∣Ω ⟩$ collapses to a single integral over the positive energy hyperboloid.

Answer

The free scalar field with mass $m$ has mode expansion $$ \phi(x) = \int \frac{d^3k}{(2\pi)^3, 2\omega_k}, \bigl[ a(\mathbf{k}) e^{-ik \cdot x} + a^(\mathbf{k}) e^{ik \cdot x} \bigr], \qquad \omega_k = \sqrt{\mathbf{k}^2 + m^2}. $$ The canonical commutation relation $[a(\mathbf{k}), a^(\mathbf{k}')] = (2\pi)^3, 2\omega_k, \delta^3(\mathbf{k} - \mathbf{k}') $an d$ a(\mathbf{k}) |\Omega\rangle = 0$ give $$ W_2(x, y) = \langle \Omega | \phi(x) \phi(y) | \Omega \rangle = \int \frac{d^3k}{(2\pi)^3, 2\omega_k}, e^{-ik \cdot (x - y)}\Big|_{k^0 = \omega_k}. $$ The Fourier transform in $ξ = x - y$ is $\hat{W}_{2} (p) = (2 π) θ (p^{0}) δ (p^{2} - m^{2})$ , the positive-energy mass-shell measure. Support lies in ${p : p^{0} \geq 0, p^{2} = m^{2}} \subset \overline{V_{+}}$ , confirming W3. The Källén-Lehmann spectral function of the free field of mass $m$ is the Dirac measure $ρ (μ^{2}) = δ (μ^{2} - m^{2})$ . Rubric: full credit for the mode-expansion computation plus the spectrum-condition check.

Exercise 4 (medium, short-answer).

Show that the locality axiom W6 forces the commutator $[ϕ (x), ϕ (y)]$ of a free real scalar field to be a $c$ -number (a multiple of the identity).

Hint

The commutator of free creation and annihilation operators is proportional to the identity. Compute it directly from the mode expansion and check that the result vanishes at spacelike separation.

Answer

Using the mode expansion from Exercise 3, $$ [\phi(x), \phi(y)] = \int \frac{d^3k}{(2\pi)^3, 2\omega_k}, \bigl( e^{-ik \cdot (x - y)} - e^{ik \cdot (x - y)} \bigr) \cdot \mathbb{1} = i, \Delta(x - y) \cdot \mathbb{1}, $$ where $Δ (ξ)$ is the Pauli-Jordan distribution $$ \Delta(\xi) = -i \int \frac{d^3k}{(2\pi)^3, 2\omega_k}, \bigl( e^{-i k \cdot \xi} - e^{i k \cdot \xi} \bigr)\Big|_{k^0 = \omega_k}. $$ Lorentz invariance of the mass-shell measure makes $Δ$ Lorentz-invariant. Direct computation shows $Δ (ξ) = 0$ when $ξ$ is spacelike, since the contour in $k^{0}$ can be closed without crossing either pole, leaving the difference of two equal contour integrals. So the commutator vanishes at spacelike separation, in line with W6, and it is a multiple of the identity at every separation. The Pauli-Jordan distribution is the unique (up to scale) Lorentz-invariant tempered distribution supported in the closed light cone and antisymmetric under $ξ \mapsto - ξ$ . Rubric: full credit for the $c$ -number identification plus the spacelike vanishing.

Exercise 5 (medium, short-answer).

State the Källén-Lehmann spectral representation of the two-point Wightman function and explain how it follows from axioms W1, W2, and W3.

Hint

Insert a complete set of energy-momentum eigenstates between the two fields. Lorentz covariance forces the spectral measure to be a function of $p^{2}$ alone, supported in $\overline{V_{+}}$ .

Answer

For a Wightman field, the Fourier transform of $W_{2}$ has the form $$ \hat{W}2(p) = \int_0^\infty d\rho(m^2), \theta(p^0), \delta(p^2 - m^2), $$ where $ρ$ is a positive tempered measure on $[0, \infty)$ called the spectral function or Källén-Lehmann weight. Insert a complete set of joint energy-momentum eigenstates $∣ p, α ⟩$ on $H$ between the fields; W2 (Poincaré covariance) of the unitary representation makes the matrix elements $⟨ Ω∣ ϕ (0) ∣ p, α ⟩$ Lorentz-covariant, and W3 (spectrum condition) confines the spectrum to $\overline{V+} $. L or e n t z in v a r ian ceo f t h e v a c uu man d o f$ \phi(0) $o na sc a l a r f i e l df or ces t h es p ec t r a l m e a s u r e t o d e p e n d o n l y o n t h e L or e n t z - in v a r ian t$ p^2 $, g i v in g t h e in t e g r a t e df or mab o v e w i t h$ \rho(m^2) = \sum_\alpha |\langle \Omega | \phi(0) | m, \alpha \rangle|^2 $a t f i x e d in v a r ian t ma ss$ m $. P os i t i v i t y o f$ \rho $i s t h e GN S p os i t i v i t y . T h e f r ee f i e l d o f ma ss$ m $ha s$ \rho(\mu^2) = \delta(\mu^2 - m^2) $; anin t er a c t in g f i e l d ha s$ \rho $s u pp or t e d o na o n e - p a r t i c l e p o l e pl u s a co n t in uu m s t a r t in g a tt h e m u l t i - p a r t i c l e t h r es h o l d . R u b r i c : f u l l cr e d i t f or t h er e p r ese n t a t i o n pl u s i d e n t i f i c a t i o n o f$ \rho$.

Exercise 6 (medium, short-answer).

Demonstrate that the uniqueness of the vacuum in W3 follows from the cluster decomposition property of the Wightman functions.

Hint

Suppose the theory has two linearly independent Poincaré-invariant unit vectors. Use them to construct a Wightman function that fails to cluster.

Answer

If there is a Poincaré-invariant unit vector $∣ Ω^{'} ⟩$ orthogonal to $∣Ω ⟩$ , the projection $P$ onto the invariant subspace is the orthogonal sum of the rank-one projections $∣Ω ⟩ ⟨ Ω∣$ and $∣ Ω^{'} ⟩ ⟨ Ω^{'} ∣$ . The connected (truncated) two-point function $W_{2}^{c} (x_{1}, x_{2}) = W_{2} (x_{1}, x_{2}) - W_{1} (x_{1}) W_{1} (x_{2})$ then receives a constant contribution from the projection onto the $∣ Ω^{'} ⟩$ direction, since $$ \langle \Omega | \phi(x_1) P \phi(x_2) | \Omega \rangle = \sum_{\Omega''} \langle \Omega | \phi(x_1) | \Omega'' \rangle \langle \Omega'' | \phi(x_2) | \Omega \rangle, $$ and at spacelike separation $x_{2} - x_{1} \to \infty$ the operator $P$ gives a nonzero limit, contradicting cluster decomposition. So vacuum uniqueness is forced by cluster decomposition; in the GNS construction of the reconstruction theorem, cluster decomposition of the $W_{n}$ is what ensures the cyclic vector $∣Ω ⟩$ is the only Poincaré-invariant unit vector up to a phase. Rubric: full credit for the spectral-projection argument and the failure-of-clustering conclusion.

Exercise 7 (hard, short-answer).

State Haag's theorem (1955) and explain why it rules out the naive interaction picture for an interacting Wightman quantum field theory.

Hint

Haag's theorem says any field that is unitarily equivalent at a single time to a free field is itself a free field. Compare equal-time CCR with the spectrum condition.

Answer

Haag's theorem. Let $ϕ$ be a Wightman field satisfying equal-time canonical commutation relations of the standard form with momentum-space dispersion fixing a mass. If there is a unitary $V$ on $H$ such that at some fixed time $t = 0$ the field $ϕ (0, x)$ is intertwined by $V$ with the free field $ϕ_{0} (0, x)$ of the same mass, and if both fields share a common Poincaré-invariant vacuum and dense cyclic domain, then $ϕ = V ϕ_{0} V^{- 1}$ as Wightman fields, so $ϕ$ itself is free.

Implication. The naive interaction picture assumes a unitary that diagonalises the free part of the Hamiltonian at one time and dresses it back into the full theory via a time-ordered exponential. Haag's theorem says such a unitary cannot connect a genuinely interacting Wightman field to a free Wightman field on the same Hilbert space with the same vacuum: the equivalence collapses both fields to the free one. Perturbation theory survives only because it operates in the formal-power-series limit and abandons the unitarity of the dressing transformation, which is exactly the price renormalisation pays. The constructive route via Osterwalder-Schrader sidesteps Haag's theorem by working on the Euclidean side, where the interaction-picture unitary obstacle does not arise. Rubric: full credit for the statement of Haag's theorem plus the interaction-picture interpretation.

Exercise 8 (hard, short-answer).

Verify all seven Wightman axioms for the free real scalar field of mass $m$ in four-dimensional Minkowski space on the bosonic Fock space $F_{s} (L^{2} (X_{m}, d λ_{m}))$ , where $X_{m}$ is the positive mass hyperboloid.

Hint

Build the field as $ϕ (f) = a^{*} (f^{+}) + a (\overline{f^{-}})$ with $f^{\pm}$ the positive- and negative-frequency mass-shell restrictions of $\hat{f}$ . Check the axioms one by one against the GNS data from the reconstruction theorem.

Answer

(W1) The Hilbert space $F_{s} (L^{2} (X_{m}, d λ_{m})) = ⨁_{n} S^{n} L^{2} (X_{m}, d λ_{m})$ is separable, since $L^{2} (X_{m}, d λ_{m})$ is separable as a Borel measure space.

(W2) The Poincaré group acts on $L^{2} (X_{m}, d λ_{m})$ by the mass- $m$ , spin-zero unitary representation (Wigner 1939). Second quantisation lifts this to a strongly continuous unitary representation $U = Γ (U_{1})$ on the Fock space, with $U_{1}$ the one-particle representation.

(W3) The translation generators $P^{μ}$ act on the one-particle space by multiplication by the four-momentum $p^{μ}$ on the mass hyperboloid $X_{m}$ , so the one-particle joint spectrum is $X_{m} \subset \overline{V_{+}}$ . Second quantisation gives joint spectrum equal to all finite sums of mass-shell momenta, which is again contained in $\overline{V_{+}}$ . The vacuum $∣Ω ⟩$ is the zero-particle state, the only Poincaré-invariant unit vector up to a phase.

(W4) Define $ϕ (f) = a^{*} (f^{+}) + a (\overline{f^{-}})$ where $f^{\pm} (p) = \hat{f} (\pm p) ∣_{p \in X_{m}}$ . The creation and annihilation operators are densely defined unbounded operators on the dense subspace $D$ of finite-particle vectors with smooth one-particle wavefunctions. Temperedness of the map $f \mapsto ϕ (f) ψ$ for $ψ \in D$ follows from the continuity of $\hat{\cdot} : S (R^{4}) \to S (R^{4})$ and the boundedness of restriction to $X_{m}$ .

(W5) Poincaré covariance of the mass-shell restriction together with the second-quantisation functor gives $U (Λ, a) ϕ (f) U (Λ, a)^{- 1} = ϕ (f_{Λ, a})$ as required.

(W6) The commutator $[ϕ (f), ϕ (g)] = i \int Δ (x - y) f (x) g (y) d x d y \cdot 1$ with the Pauli-Jordan distribution $Δ$ vanishing at spacelike separation (Exercise 4).

(W7) Polynomials in $a^{*} (f^{+})$ acting on $∣Ω ⟩$ produce every $n$ -particle state, and varying $f$ over $S (R^{4})$ produces a dense subspace of $L^{2} (X_{m}, d λ_{m})$ in each one-particle slot, so the cyclic span is dense in $F_{s}$ .

All seven axioms hold, confirming the free scalar field as a Wightman quantum field theory. Rubric: full credit for the seven-fold verification with named justifications.

Advanced results Master

Theorem (PCT theorem; Lüders 1954 ^{[Lüders 1954]}, Pauli 1955, Jost 1957). Let $(H, U, ∣Ω ⟩, {ϕ_{j}})$ be a Wightman quantum field theory. There exists an antiunitary operator $Θ : H \to H$ satisfying $Θ∣Ω ⟩ = ∣Ω ⟩$ and $$ \Theta, \phi_j(x), \Theta^{-1} = (-1)^{J_j}, \overline{\phi_j(-x)}^*, $$ where $J_{j}$ is the spinorial rank of $ϕ_{j}$ . Equivalently, the Wightman functions are invariant under simultaneous reflection of all spacetime arguments, complex conjugation, and Hermitian conjugation.

The proof proceeds by analytically continuing the Wightman functions $W_{n} (x_{1}, \dots, x_{n})$ in the difference variables $ξ_{k} = x_{k} - x_{k + 1}$ . The spectrum condition forces $\hat{W}_{n}$ to be supported in $\overline{V_{+}}^{n - 1}$ , so the Laplace transform of $\hat{W}_{n}$ is analytic on the forward tube $T^{-} = {(ζ_{1}, \dots, ζ_{n - 1}) : Im ζ_{k} \in - V_{+}}$ . Lorentz invariance extends the domain of analyticity from the forward tube to the Bargmann-Hall-Wightman extended tube $T_{n - 1}^{'}$ , which contains real Jost points — real Lorentz-frames with all $ξ_{k}$ spacelike. At a Jost point, all permutations of the $x_{k}$ give equal $W_{n}$ values by locality, and analytic continuation through the extended tube transports this equality to the original arguments with reversed sign and complex conjugation. The result is the PCT statement.

Theorem (Spin-statistics theorem; Pauli 1940 ^{[Pauli 1940]}, Burgoyne 1958 ^{[Burgoyne 1958]}). In a Wightman quantum field theory, a field $ϕ_{j}$ of integer spin $J_{j} \in Z$ satisfies the locality commutator $[ϕ_{j} (f), ϕ_{j} (g)] = 0$ at spacelike separation, while a field of half-integer spin $J_{j} \in Z + 1/2$ satisfies the anticommutator ${ϕ_{j} (f), ϕ_{j} (g)} = 0$ . The reversed assignment — half-integer-spin fields commuting at spacelike separation, integer-spin fields anticommuting — forces the field to vanish identically.

The argument: the analytic continuation of $W_{2}$ through the extended tube, combined with locality at a Jost point and positivity of $⟨ Ω∣ ϕ (f) ϕ (f)^{*} ∣Ω ⟩$ , produces an analytic function on the tube with a specific sign behaviour under the Lorentz spinorial parity. Integer spin gives one sign, half-integer spin the other. Mismatched statistics force the spectral function $ρ$ in the Källén-Lehmann representation to be either zero or negative — the first option means the field is zero, and the second contradicts positivity. So the spin-statistics pairing is the only one consistent with all six structural axioms together.

Theorem (Haag's theorem; Haag 1955 ^{[Haag 1955]}). Let $ϕ$ be a Wightman field with the equal-time canonical commutation relations of a free scalar of mass $m$ . If $ϕ$ is unitarily equivalent at $t = 0$ to the free field $ϕ_{0}$ of the same mass on a common Poincaré-invariant vacuum $∣Ω ⟩$ , then $ϕ = ϕ_{0}$ as Wightman fields.

The proof identifies the equal-time CCR algebra of $ϕ$ with that of $ϕ_{0}$ via the unitary $V$ , then notes that Lorentz invariance of the vacuum and the spectrum condition extend the equality from $t = 0$ to every time. Since both fields share the same vacuum and the same Lorentz-invariant Wightman functions, the reconstruction theorem makes them unitarily equivalent as Wightman fields. The naive interaction picture proposes precisely such a unitary; Haag's theorem rules it out at the level of bona fide Wightman fields, and perturbation theory survives only by working in a formal-power-series ring where the unitary equivalence is replaced by a formal expansion.

Theorem (Borchers algebra reformulation; Borchers 1962 ^{[Borchers 1962]}). A Wightman quantum field theory is the same datum as a continuous, Poincaré-invariant, positive, locality-symmetric linear functional $ω$ on the Borchers tensor algebra $B = ⨁_{n} S (R^{4 n})$ , with $ω (1) = 1$ , such that the GNS Hilbert space carries a unique Poincaré-invariant unit vector and the spectrum of the GNS translation generators lies in $\overline{V_{+}}$ .

Borchers's reformulation turns the seven axioms into one continuous functional on a single algebra. The GNS construction of the reconstruction theorem is then the assertion that this functional is the same data as the Wightman quantum field theory. The advantage is conceptual: one functional is easier to vary, to deform, and to compare with other functionals than seven separate axioms on a tower of objects.

Theorem (Haag-Kastler algebraic axioms; Haag-Kastler 1964 ^{[Haag-Kastler 1964]}). Replace the field-by-field Wightman structure with a net of local von Neumann algebras $O \mapsto A (O)$ indexed by open bounded subsets of $R^{1, 3}$ , satisfying isotony, locality (algebras at spacelike-separated regions commute), Poincaré covariance, and the spectrum condition on the universal representation. Every Wightman quantum field theory canonically generates such a net by setting $A (O) = \overline{poly {ϕ_{j} (f) : supp f \subseteq O}}^{''}$ .

The algebraic axioms remove the choice of field generators, the choice of a dense domain, and the equal-time canonical commutation relations from the description. What remains is the bare net of local observables, which suffices for most structural results about scattering, superselection sectors, and the analysis of charged states (the DHR theory of Doplicher-Haag-Roberts 1969-1974). Many results that are intricate in the Wightman framework — Reeh-Schlieder, modular theory, the type-III nature of local algebras — are cleanest in the Haag-Kastler net.

Theorem (Reeh-Schlieder; Reeh-Schlieder 1961). Let $A (O)$ be the local algebra of a Wightman quantum field theory on an open bounded region $O$ . The vacuum $∣Ω ⟩$ is cyclic for $A (O)$ — the closure $\overline{A (O) ∣Ω ⟩}$ is all of $H$ — and it is also separating: no nonzero element of $A (O)$ annihilates $∣Ω ⟩$ .

Reeh-Schlieder is the deepest immediate consequence of the spectrum condition. Polynomials in fields smeared in an arbitrarily small open region already sweep out the entire Hilbert space, by analytic continuation of vacuum expectation values from the global cyclic span (W7) to the local one. The result is highly counterintuitive: an experimentalist in a small laboratory could, in principle, prepare any global state of the universe — though the field operators involved would be highly unphysical. The theorem warns that the localised observables of a Wightman theory carry nonlocal correlations, a fact that has consequences for entanglement structure and for the modular theory of Tomita-Takesaki.

Synthesis. The Wightman axioms are the foundational reason quantum field theory is a meaningful subject independent of any perturbative recipe: they identify a relativistic quantum theory with a sequence of vacuum expectation values, and the reconstruction theorem is exactly the assertion that this identification is bijective up to unitary equivalence. The central insight is that positivity, Poincaré invariance, the spectrum condition, locality, and cluster decomposition together force PCT, spin-statistics, Reeh-Schlieder, and the Borchers-algebra reformulation; each of these is dual to one of the axioms in a precise sense, and the package is rigid enough that altering one axiom would alter the whole list. This is exactly the same rigidity that appears again in 08.10.05 when the Feynman propagator's analyticity properties — the $i ϵ$ prescription, the pole structure, the contour-integral representation — are recognised as the two-point Wightman function's reflection of the spectrum condition. Putting these together, the seven Wightman axioms, the Borchers algebra functional, the Haag-Kastler net of local algebras, and the Osterwalder-Schrader Euclidean axioms form one framework with four equivalent presentations, and the bridge between any two is a single theorem: GNS for Wightman to Borchers, generation-of-local-algebras for Wightman to Haag-Kastler, and analytic continuation for Wightman to Osterwalder-Schrader. The foundational reason this works is that each equivalence preserves positivity, the appropriate invariance, and the appropriate notion of locality; the central insight is that one set of structural conditions can be packaged in many algebraic shells without losing any content.

The Wightman framework also identifies the precise obstruction to interacting QFT in four dimensions: Haag's theorem rules out the naive interaction picture, the Källén-Lehmann representation forces the spectral function of an interacting field to have a continuum threshold above the one-particle mass, and the constructive program of Glimm-Jaffe and Osterwalder-Schrader produces interacting Wightman theories in two and three spacetime dimensions but no four-dimensional example is known. Aizenman-Duminil-Copin 2021 closed the four-dimensional $ϕ^{4}$ chapter by proving triviality of the lattice continuum limit, so the candidate constructive route in four dimensions has been eliminated for that particular Lagrangian. The bridge is that the obstruction is not in the axioms themselves but in the spectrum of relativistic, locally interacting, four-dimensional theories satisfying them, an obstruction whose precise statement appears again in the Yang-Mills mass-gap problem.

Full proof set Master

Theorem (Wightman-Gårding reconstruction), proof. Given in the Intermediate-tier section: the GNS construction applied to the Borchers tensor algebra produces a Hilbert space, a vacuum, a Poincaré representation, and fields with the seven axioms holding by construction. Uniqueness follows from cyclicity (W7) and equality of matrix elements via the $W_{n}$ . The seven-step verification in the body of the proof covers W1 through W7 in sequence. $□$

Proposition (Wick rotation of two-point function). The two-point Wightman function $W_{2} (x, y)$ analytically continues from the forward tube $Im (x - y) \in - V_{+}$ to the Schwinger function $S_{2} (x_{E}, y_{E})$ at Euclidean separation $x_{E} - y_{E} = (i τ, x - y)$ , and the continuation identifies the Källén-Lehmann spectral function $ρ$ with the Euclidean spectral measure of the Schwinger two-point function.

Proof. The spectrum condition makes $\hat{W}_{2} (p)$ supported in $\overline{V_{+}}$ , so $W_{2} (x - y) = \int e^{- i p \cdot (x - y)} d \hat{W}_{2} (p)$ admits an analytic continuation to $x - y \to x_{E} - y_{E}$ with $Im (x - y)^{0} = - τ < 0$ , given by $$ S_2(\tau, \mathbf{x} - \mathbf{y}) = \int e^{-p^0 \tau - i \mathbf{p} \cdot (\mathbf{x} - \mathbf{y})}, d\hat{W}_2(p), $$ where the exponent has positive real part exactly because $p^{0} \geq 0$ on the support of $\hat{W}_{2}$ and $τ > 0$ . The integral converges absolutely, defines an analytic function in $τ$ , and equals the Euclidean two-point function. Under the Källén-Lehmann decomposition $\hat{W}_{2} = \int d ρ (m^{2}) θ (p^{0}) δ (p^{2} - m^{2})$ , the Schwinger function becomes $S_{2} (τ, r) = \int d ρ (m^{2}) G_{m} (τ, r)$ with $G_{m}$ the Euclidean Klein-Gordon propagator of mass $m$ , so the spectral function is the same on both sides. $□$

Proposition (Reeh-Schlieder cyclicity). Let $O \subseteq R^{1, 3}$ be an open bounded region. The vacuum $∣Ω ⟩$ is cyclic for the local algebra $A (O)$ .

Proof. Let $ψ \in H$ be orthogonal to $A (O) ∣Ω ⟩$ . For any test functions $f_{1}, \dots, f_{n}$ with supports in $O$ , $$ \langle \psi | \phi(f_1) \cdots \phi(f_n) | \Omega \rangle = 0. $$ Translate the field operators by $a \in R^{1, 3}$ : $ϕ (f_{k})$ becomes $U (1, a) ϕ (f_{k}) U (1, a)^{- 1} = ϕ (f_{k} (\cdot - a))$ . For sufficiently small $a$ this remains supported in $O$ , so the matrix element above continues to vanish. By Stone's theorem the dependence on $a$ is analytic on the forward tube — the spectrum condition makes $U (1, a) = e^{- i P \cdot a}$ analytic in $Im a \in - V_{+}$ . By the edge-of-the-wedge theorem, analytic continuation through the tube gives a function vanishing on an open set of $R^{1, 3}$ , hence vanishing identically. Translating the $f_{k}$ to arbitrary locations and applying global cyclicity W7 produces a vector $ψ$ orthogonal to every vector in the cyclic span, hence $ψ = 0$ . So $A (O) ∣Ω ⟩$ is dense in $H$ . $□$

Proposition (spin-statistics for the free scalar). The free real scalar field of mass $m$ satisfies the locality commutator $[ϕ (f), ϕ (g)] = 0$ at spacelike separation, in line with its spin-zero (integer-spin) assignment.

Proof. The commutator is the Pauli-Jordan distribution $i Δ$ smeared against the two test functions (Exercise 4). The mass-shell measure is Lorentz-invariant; analytic continuation of $Δ$ to the forward tube and back through the extended tube to a Jost point gives $Δ (- ξ) = - Δ (ξ)$ , an odd function under spacetime reflection. At spacelike $ξ$ the symmetry $ξ \mapsto - ξ$ is in the connected component of the proper Lorentz group, so $Δ (ξ) = Δ (- ξ)$ ; combined with odd parity this forces $Δ (ξ) = 0$ at spacelike separation. Reversing the sign assignment — anticommuting integer-spin fields — would replace the commutator with an anticommutator ${ϕ (f), ϕ (g)}$ equal to the two-point function $W_{2}$ symmetrised in its arguments. At a Jost point with $f$ supported in a small neighbourhood, $W_{2} (f, f) = ⟨ Ω∣ ϕ (f) ϕ (f) ∣Ω ⟩ > 0$ by positivity (W3, W4), so the symmetrised sum is strictly positive, never zero. The field would have to vanish identically. $□$

Theorem (PCT theorem), stated without proof — see Streater-Wightman Ch. 4 ^{[Streater-Wightman 1964]} and Jost 1957. The full proof uses the Bargmann-Hall-Wightman analytic-continuation theorem to extend the Wightman functions from the forward tube to the extended tube, applies locality at Jost points, and identifies the resulting symmetry as PCT. The argument is technical but completely elementary in structure; it requires no input beyond the seven axioms.

Theorem (spin-statistics theorem), proof outline. Combine the Källén-Lehmann representation of the two-point function with analytic continuation through the extended tube and positivity. For half-integer spin, the spinorial parity contributes a factor $(- 1)^{2 J}$ in the analytic continuation that flips the sign of the integrated spectral function relative to the integer-spin case. A commutator at spacelike separation pairs with the integer-spin sign and an anticommutator with the half-integer-spin sign; the reversed pairing forces the spectral function to be non-positive, which combined with the GNS positivity (W3, W4) makes it zero, hence the field vanishes. The proof in this form is Burgoyne 1958 ^{[Burgoyne 1958]}. $□$

Theorem (Haag's theorem), proof outline. A unitary equivalence at $t = 0$ between $ϕ$ and the free field $ϕ_{0}$ , combined with shared Lorentz invariance of the vacuum, propagates to a unitary equivalence at every time. Both fields satisfy the same equal-time CCR, share the same vacuum, and produce the same Wightman functions $W_{n}$ . The reconstruction theorem then makes them unitarily equivalent as Wightman fields, hence equal up to unitary relabelling. The argument fails when one drops the requirement that the vacuum is Lorentz-invariant (one then enters the realm of Bogoliubov-transformed Fock representations), or when one works in a formal-power-series ring rather than on a fixed Hilbert space. $□$

Connections Master

Bosonic Fock space and second quantisation 08.10.01. The bosonic Fock space $F_{s} (L^{2} (X_{m}, d λ_{m}))$ over the mass-shell one-particle Hilbert space is the explicit worked example of a Wightman quantum field theory; W1 names this space, and the second-quantisation functor packages the Poincaré representation of W2 and the spectrum condition of W3 as the canonical Fock-space lift of the mass- $m$ Wigner representation. The free field constructed in 08.10.01 is the running model the seven axioms are calibrated against.
Wick's theorem for operator products 08.10.04. Wick's theorem is the structural property of the free Wightman field that makes every $n$ -point Wightman function a sum over complete pairings of two-point functions. Under reconstruction, the free Wightman theory is characterised by its two-point function alone, and Wick's theorem is the assertion that the higher Wightman functions are determined by Gaussian moment combinatorics. For an interacting Wightman field, Wick's theorem fails and the higher $W_{n}$ carry independent information encoded by truncated (connected) Wightman functions.
Feynman propagator and the contour-integral representation 08.10.05. The Feynman propagator is the time-ordered combination of the two Wightman two-point functions, $Δ_{F} (x - y) = θ (x^{0} - y^{0}) W_{2} (x, y) + θ (y^{0} - x^{0}) W_{2} (y, x)$ . The $i ϵ$ prescription, the contour-integral representation, and the pole structure all reflect the spectrum condition W3 of the underlying Wightman theory; the propagator is the analytic incarnation of W3 at the level of two-point distributions, and the Källén-Lehmann representation is the general-spectral-measure extension.
Dirac equation and relativistic spin 12.11.01. Half-integer-spin Wightman fields are quantised Dirac fields, and the spin-statistics theorem identifies them as the unique anticommuting Wightman fields compatible with positivity of energy and microcausality. The Lorentz representation $S (Λ)$ in axiom W5 specialises to the $(1/2, 0) \oplus (0, 1/2)$ representation of $SL (2, C)$ on the Dirac spinor space, and Pauli-Lüders PCT identifies the antiunitary $Θ$ with the product of charge conjugation, parity, and time reversal acting on Dirac fields.
Hilbert space 02.11.08 and unbounded self-adjoint operators 02.11.03. The seven axioms operate on the analytic substrate of a separable Hilbert space carrying a continuous unitary representation of the Poincaré group, with the energy-momentum operators $P^{μ}$ as commuting unbounded self-adjoint operators satisfying the spectrum condition. Stone's theorem applied to the translation subgroup produces $P^{μ}$ from the unitary representation, and the spectral theorem gives the joint spectral measure whose support is constrained to $\overline{V_{+}}$ by W3.

Historical & philosophical context Master

Wightman introduced the axiomatic framework that bears his name in Phys. Rev. 101, 860 (1956) ^{[Wightman 1956]}, proposing to characterise a relativistic quantum field theory by the sequence of vacuum expectation values of products of fields rather than by a Lagrangian. The motivating context was the post-renormalisation impasse of quantum electrodynamics: Tomonaga, Schwinger, Feynman and Dyson had produced spectacular numerical agreement with experiment by 1948-1950, but the perturbative series itself was understood to be asymptotic, not convergent, and Haag's 1955 theorem ^{[Haag 1955]} had ruled out the naive interaction-picture unitary equivalence between interacting and free fields. Axiomatic QFT was meant to identify what a quantum field theory is before asking how to compute with one. Wightman's paper laid out positivity, Poincaré invariance, the spectrum condition, locality, and cluster decomposition as the minimal conditions on the vacuum expectation values, and the Wightman-Gårding theorem of Ark. Fys. 28, 129 (1964) ^{[Wightman-Gårding 1964]} completed the picture by showing that these conditions are also sufficient: a sequence of distributions with these properties reconstructs a unique quantum field theory.

The Streater-Wightman monograph PCT, Spin and Statistics, and All That (Benjamin, 1964; Princeton Landmarks reissue, 2000) ^{[Streater-Wightman 1964]} presented the framework in its mature form together with the two main structural theorems: PCT (Pauli-Lüders-Schwinger; Lüders Mat.-Fys. Medd. 28, no. 5 (1954) ^{[Lüders 1954]}; Pauli 1955; Schwinger Phys. Rev. 82, 914 (1951) ^{[Schwinger 1951]}) and spin-statistics (Pauli Phys. Rev. 58, 716 (1940) ^{[Pauli 1940]}; Burgoyne Nuovo Cim. 8, 607 (1958) ^{[Burgoyne 1958]}). Pauli's 1940 argument predates the axiomatic framework; it relied on Lorentz invariance and positivity of energy applied to a free-field quantisation. Burgoyne 1958 rewrote the same argument inside the Wightman axioms, isolating exactly which structural inputs were doing the work. Jost's 1957 paper introduced the analytic-continuation route through the Bargmann-Hall-Wightman extended tube that produced the cleanest known proof of PCT.

Borchers reformulated the framework in Nuovo Cim. 24, 214 (1962) ^{[Borchers 1962]} by packaging the seven axioms as one continuous positive Poincaré-invariant functional on the tensor algebra $⨁_{n} S (R^{4 n})$ , now called the Borchers algebra. Haag and Kastler in J. Math. Phys. 5, 848 (1964) ^{[Haag-Kastler 1964]} proposed the algebraic-QFT alternative: a net of local von Neumann algebras over open regions, with isotony, locality, Poincaré covariance, and the spectrum condition on the universal representation. Algebraic QFT became the dominant framework for structural results from the late 1960s onward (Doplicher-Haag-Roberts 1969-1974 on superselection sectors; Bisognano-Wichmann 1975 on modular theory of wedge algebras), while constructive QFT pursued the Wightman side through the Osterwalder-Schrader theorem (Osterwalder-Schrader Commun. Math. Phys. 31, 83 (1973); 42, 281 (1975)) and the Glimm-Jaffe program (Glimm-Jaffe Quantum Physics, 2nd ed., Springer 1987) that built interacting Wightman theories in two and three spacetime dimensions. Bogoliubov, Logunov, Oksak and Todorov collected the analytic side in Introduction to Axiomatic Quantum Field Theory (Benjamin 1975, expanded Kluwer 1990), with extensive treatment of dispersion relations and the edge-of-the-wedge consequences of the spectrum condition.

Bibliography Master

@article{Wightman1956,
  author  = {Wightman, A. S.},
  title   = {Quantum field theory in terms of vacuum expectation values},
  journal = {Phys. Rev.},
  volume  = {101},
  year    = {1956},
  pages   = {860--866}
}

@book{StreaterWightman1964,
  author    = {Streater, R. F. and Wightman, A. S.},
  title     = {PCT, Spin and Statistics, and All That},
  publisher = {W. A. Benjamin},
  year      = {1964},
  note      = {Princeton Landmarks in Physics reissue, Princeton University Press, 2000}
}

@article{WightmanGarding1964,
  author  = {Wightman, A. S. and G{\aa}rding, L.},
  title   = {Fields as operator-valued distributions in relativistic quantum theory},
  journal = {Ark. Fys.},
  volume  = {28},
  year    = {1964},
  pages   = {129--184}
}

@article{Borchers1962,
  author  = {Borchers, H.-J.},
  title   = {On structure of the algebra of field operators},
  journal = {Nuovo Cim.},
  volume  = {24},
  year    = {1962},
  pages   = {214--236}
}

@article{Haag1955,
  author  = {Haag, R.},
  title   = {On quantum field theories},
  journal = {Kgl. Danske Vidensk. Selsk. Mat.-Fys. Medd.},
  volume  = {29},
  number  = {12},
  year    = {1955}
}

@article{HaagKastler1964,
  author  = {Haag, R. and Kastler, D.},
  title   = {An algebraic approach to quantum field theory},
  journal = {J. Math. Phys.},
  volume  = {5},
  year    = {1964},
  pages   = {848--861}
}

@article{Pauli1940,
  author  = {Pauli, W.},
  title   = {The connection between spin and statistics},
  journal = {Phys. Rev.},
  volume  = {58},
  year    = {1940},
  pages   = {716--722}
}

@article{Luders1954,
  author  = {L{\"u}ders, G.},
  title   = {On the equivalence of invariance under time reversal and under particle-antiparticle conjugation for relativistic field theories},
  journal = {Kgl. Danske Vidensk. Selsk. Mat.-Fys. Medd.},
  volume  = {28},
  number  = {5},
  year    = {1954}
}

@article{Schwinger1951,
  author  = {Schwinger, J.},
  title   = {The theory of quantized fields. I},
  journal = {Phys. Rev.},
  volume  = {82},
  year    = {1951},
  pages   = {914--927}
}

@article{Burgoyne1958,
  author  = {Burgoyne, N.},
  title   = {On the connection of spin with statistics},
  journal = {Nuovo Cim.},
  volume  = {8},
  year    = {1958},
  pages   = {607--609}
}

@article{OsterwalderSchrader1973,
  author  = {Osterwalder, K. and Schrader, R.},
  title   = {Axioms for {E}uclidean {G}reen's functions},
  journal = {Commun. Math. Phys.},
  volume  = {31},
  year    = {1973},
  pages   = {83--112}
}

@article{OsterwalderSchrader1975,
  author  = {Osterwalder, K. and Schrader, R.},
  title   = {Axioms for {E}uclidean {G}reen's functions. {II}},
  journal = {Commun. Math. Phys.},
  volume  = {42},
  year    = {1975},
  pages   = {281--305}
}

@book{BogoliubovLogunovOksakTodorov1990,
  author    = {Bogoliubov, N. N. and Logunov, A. A. and Oksak, A. I. and Todorov, I. T.},
  title     = {General Principles of Quantum Field Theory},
  publisher = {Kluwer Academic Publishers},
  year      = {1990}
}

@book{GlimmJaffe1987,
  author    = {Glimm, J. and Jaffe, A.},
  title     = {Quantum Physics: A Functional Integral Point of View},
  publisher = {Springer-Verlag},
  edition   = {2},
  year      = {1987}
}

@article{ReehSchlieder1961,
  author  = {Reeh, H. and Schlieder, S.},
  title   = {Bemerkungen zur Unit{\"a}r{\"a}quivalenz von Lorentzinvarianten Feldern},
  journal = {Nuovo Cim.},
  volume  = {22},
  year    = {1961},
  pages   = {1051--1068}
}

@article{ParisiWu1981,
  author  = {Parisi, G. and Wu, Y.-S.},
  title   = {Perturbation theory without gauge fixing},
  journal = {Sci. Sin.},
  volume  = {24},
  year    = {1981},
  pages   = {483--496}
}

Prerequisites

02.11.08
02.11.03
08.10.01
08.10.04
08.10.05

Tier anchors

beginner: Chatterjee, *Introduction to Quantum Field Theory for Mathematicians* (Stanford lecture notes, 2022), Lectures 22-23 (axiomatic setting and the seven Wightman conditions); Tong, *Quantum Field Theory* (DAMTP lecture notes, 2007), §2 (canonical quantisation and the free scalar field as a Lorentz-covariant operator-valued distribution)
intermediate: Streater & Wightman, *PCT, Spin and Statistics, and All That* (Benjamin, 1964; Princeton Landmarks reissue, 2000), Chs. 2-3 (the axioms, Wightman functions, the reconstruction theorem); Bogoliubov, Logunov, Oksak & Todorov, *General Principles of Quantum Field Theory* (Kluwer, 1990), Chs. 7-10; Folland, *Quantum Field Theory: A Tourist Guide for Mathematicians* (AMS Math. Surveys 149, 2008), Ch. 7
master: Wightman, *Phys. Rev.* 101, 860 (1956) (the original Wightman-function paper); Wightman & Gårding, *Ark. Fys.* 28, 129 (1964) (the reconstruction theorem in its definitive form); Streater-Wightman, *PCT, Spin and Statistics, and All That* (1964); Borchers, *Nuovo Cim.* 24, 214 (1962) (the Borchers tensor algebra approach); Haag, *Kgl. Danske Vidensk. Selsk. Mat.-Fys. Medd.* 29, 12 (1955) (the no-go theorem on the interaction picture); Haag & Kastler, *J. Math. Phys.* 5, 848 (1964) (algebraic axioms as the continuous-spectrum alternative); Glimm & Jaffe, *Quantum Physics: A Functional Integral Point of View*, 2nd ed. (Springer, 1987), Ch. 6

References

Wightman, A. S. — Quantum field theory in terms of vacuum expectation values · Phys. Rev. 101, 860 (1956) — the original paper formulating axiomatic QFT in terms of vacuum expectation values of products of fields and identifying the positivity, Poincaré-invariance, spectrum and locality conditions that the n-point distributions must satisfy
Streater, R. F. & Wightman, A. S. — PCT, Spin and Statistics, and All That · W. A. Benjamin (1964); Princeton Landmarks in Physics reissue (2000). Ch. 2 (the axioms), Ch. 3 (Wightman functions and the reconstruction theorem), Ch. 4 (PCT and spin-statistics)
Wightman, A. S. & Gårding, L. — Fields as operator-valued distributions in relativistic quantum theory · Ark. Fys. 28, 129 (1964) — the definitive form of the reconstruction theorem; from a sequence of tempered distributions on $\mathbb{R}^{4n}$ satisfying positivity, Poincaré invariance, spectrum, locality and cluster decomposition, one reconstructs a unique Wightman quantum field theory up to unitary equivalence
Borchers, H.-J. — On structure of the algebra of field operators · Nuovo Cim. 24, 214 (1962) — the Borchers tensor algebra $\mathcal{B} = \bigoplus_n \mathcal{S}(\mathbb{R}^{4n})$ as the universal target for the field map; the axioms become a continuous positive Poincaré-invariant functional on $\mathcal{B}$ vanishing on the locality ideal
Haag, R. — On quantum field theories · Kgl. Danske Vidensk. Selsk. Mat.-Fys. Medd. 29, no. 12 (1955) — Haag's theorem: any field unitarily equivalent at one time to a free field is itself free; the obstruction that rules out the naive interaction picture in Wightman QFT
Haag, R. & Kastler, D. — An algebraic approach to quantum field theory · J. Math. Phys. 5, 848 (1964) — the net of local von Neumann algebras $\mathcal{O} \mapsto \mathcal{A}(\mathcal{O})$ as the continuous-spectrum / model-independent reformulation of the Wightman axioms
Pauli, W. — The connection between spin and statistics · Phys. Rev. 58, 716 (1940) — the original spin-statistics paper; integer-spin fields commute at spacelike separation, half-integer-spin fields anticommute, given relativistic invariance and positivity of energy
Lüders, G. — On the equivalence of invariance under time reversal and under particle-antiparticle conjugation for relativistic field theories · Kgl. Danske Vidensk. Selsk. Mat.-Fys. Medd. 28, no. 5 (1954) — the PCT theorem in its first general form
Schwinger, J. — The theory of quantized fields. I · Phys. Rev. 82, 914 (1951) — Schwinger's independent route to spin-statistics and PCT from a Lorentz-invariant action principle; the Lagrangian-side counterpart of the Wightman axiomatic-side argument
Burgoyne, N. — On the connection of spin with statistics · Nuovo Cim. 8, 607 (1958) — the proof of the spin-statistics theorem within the Wightman axioms in essentially its modern form, cleaning up the Pauli-Lüders argument
Bogoliubov, N. N., Logunov, A. A. & Todorov, I. T. — Introduction to Axiomatic Quantum Field Theory · W. A. Benjamin (1975) — alternative monograph treatment of the Wightman framework with extensive coverage of the analyticity, edge-of-the-wedge and dispersion-relation consequences of the axioms
Glimm, J. & Jaffe, A. — Quantum Physics: A Functional Integral Point of View, 2nd ed. · Springer (1987) — Ch. 6 verifies the Wightman axioms for the free scalar field on Fock space and outlines the Osterwalder-Schrader Euclidean reformulation that gives the only known constructive route to interacting Wightman theories in two and three spacetime dimensions
Chatterjee, S. — Introduction to Quantum Field Theory for Mathematicians (Stanford lecture notes, 2022) · Lectures 22-23 — the seven Wightman axioms with the free scalar field as the running example; spectrum condition, microcausality, and cyclicity of the vacuum
Parisi, G. & Wu, Y.-S. — Perturbation theory without gauge fixing · Sci. Sin. 24, 483 (1981) — stochastic quantisation: the Langevin route to the Euclidean Schwinger functions whose Osterwalder-Schrader reconstruction is a Wightman theory
Osterwalder, K. & Schrader, R. — Axioms for Euclidean Green's functions · Commun. Math. Phys. 31, 83 (1973); 42, 281 (1975) — the equivalence between Wightman axioms in Minkowski signature and the Osterwalder-Schrader axioms on the Euclidean Schwinger functions, providing the bridge to constructive QFT and to stat-mech-style Euclidean lattice approaches

Estimated time

beginner: 25m
intermediate: 60m
master: 100m