Remarks and Open Issues

A collection of foundational results, caveats, and alternative formulations that complement the Wightman-style postulates of QFT.

Wightman Reconstruction Theorem

From a set of vacuum expectation values

$W_{n} (x_{1}, \dots, x_{n}) = ⟨ 0∣ ϕ (x_{1}) \dots ϕ (x_{n}) ∣0 ⟩$

satisfying Poincaré covariance, the spectrum condition, locality, hermiticity, and positivity, one can reconstruct the entire QFT — Hilbert space, fields, and vacuum — uniquely up to unitary equivalence. This makes the Wightman functions $W_{n}$ the fundamental data of a Wightman QFT.

Haag's Theorem

The interaction picture, used heuristically in textbook perturbation theory, does not strictly exist in interacting QFT: the free and interacting fields cannot act on the same Hilbert space (they are unitarily inequivalent). Perturbative QFT must therefore be understood as a formal expansion, justified by renormalization, rather than as a strict consequence of the Wightman axioms.

Gauge Theories

For theories with local gauge symmetry (e.g. QED, QCD) the postulates above must be supplemented with a gauge-fixing procedure — Gupta–Bleuler, Faddeev–Popov, BRST quantization, etc. The physical Hilbert space is then a quotient or subspace of the full state space, defined by gauge-invariance conditions.

In particular, gauge fields like the photon $A_{μ}$ do not satisfy strict Wightman positivity in covariant gauges (unphysical polarizations have negative norm); they only become consistent on the physical subspace.

Status of Rigorous Construction

No interacting QFT in four spacetime dimensions has been rigorously shown to satisfy all Wightman axioms. Constructive QFT has succeeded in lower dimensions:

$d = 2$ : $ϕ_{2}^{4}$ , sine-Gordon, Thirring, ...
$d = 3$ : $ϕ_{3}^{4}$ , Yukawa $_{3}$ , ...

The four-dimensional case — including QED, QCD, and the entire Standard Model — remains an open problem. Establishing the existence of a non-trivial Yang–Mills theory in $d = 4$ with a mass gap is one of the Millennium Prize Problems.

Algebraic QFT (Haag–Kastler)

An alternative axiomatic framework takes the primary objects to be local algebras of observables $A (O)$ associated with bounded spacetime regions $O$ , rather than fields. The postulates are then phrased as conditions on the net of algebras $O \mapsto A (O)$ :

Isotony: $O_{1} \subset O_{2} ⟹ A (O_{1}) \subset A (O_{2})$ .
Locality: $A (O_{1})$ and $A (O_{2})$ commute when $O_{1}$ , $O_{2}$ are spacelike-separated.
Covariance: the Poincaré group acts by automorphisms compatible with the net.
Spectrum condition: as in the Wightman framework.

Algebraic QFT clarifies several conceptual issues — superselection sectors, charge structure, particle statistics in low dimensions (anyons) — that are awkward in the field-based formulation.

Measurement and Collapse: Inherited, Not Derived

QFT is a quantum theory, so it inherits the entire QM measurement framework — but does it derive the Born rule (Postulate 3) and collapse (Postulate 4) of QM? Strictly: no. They are presupposed by QFT, not produced by it.

How Each QM Postulate Fares in QFT

QM postulate	Status in QFT
1 — State space	Inherited; generalized to relativistic state space (QFT Postulate 1)
2 — Observables	Inherited; observables are local field operators / algebras (QFT Postulate 4)
3 — Born rule	Inherited as primitive. The form of the rule is partially derived (Gleason's theorem); that probabilities exist is postulated.
4 — Collapse	Inherited as primitive. Decoherence partially explains the appearance of collapse but not the selection of a unique outcome.
5 — Schrödinger evolution	Generalized to dynamics from a local action (QFT Postulate 9); Lorentz-covariantized.
6 — Composite systems (tensor product)	Inherited; QFT Hilbert spaces are tensor products of per-species Fock spaces.
7 — Symmetrization	Promoted to a theorem — the spin–statistics theorem.

So QFT promotes one QM postulate (symmetrization → spin–statistics theorem), generalizes others (state space, observables, evolution), and inherits unchanged the measurement and collapse postulates.

Where Measurement Appears in QFT Practice

In practice, "measurement" in QFT means three things, all of which use the Born rule and none of which derive it:

S-matrix elements $⟨ f, out ∣ i, in ⟩$ , interpreted as transition probabilities via $∣ M_{f i} ∣^{2}$ .
Cross sections $d σ$ and decay rates $d Γ$ , derived from $∣ M ∣^{2}$ by standard kinematic factors.
Vacuum expectation values $⟨ 0∣ \hat{A} ∣0 ⟩$ and correlation functions $⟨ 0∣ T \hat{ϕ} (x_{1}) \dots \hat{ϕ} (x_{n}) ∣0 ⟩$ , interpreted as expectation values via $⟨ \hat{A} ⟩ = ⟨ ψ ∣ \hat{A} ∣ ψ ⟩$ .

The collapse postulate is rarely invoked explicitly in standard QFT because S-matrix calculations only ask for initial and final state probabilities, never for the post-measurement state. So one can do a great deal of QFT without ever using collapse — but it lurks whenever a post-measurement state is required.

Born Rule for Scattering

The "Born rule for scattering" — invoked in QFT/cross-sections.md as a foundational postulate behind the cross-section formula — is literally the QM Born rule applied to a particular kind of measurement. The same name with different-looking formulas reflects bookkeeping, not new physics.

Specialization, not a new postulate

The general QM Born rule for transitions: a system in $∣ ψ_{i} ⟩$ at $t_{0}$ evolves under unitary $\hat{U} (t, t_{0})$ , and the probability of finding it in $∣ ψ_{f} ⟩$ at $t$ is

$P_{i \to f} = ∣ ⟨ ψ_{f} ∣ \hat{U} (t, t_{0}) ∣ ψ_{i} ⟩ ∣^{2} .$

For scattering, take:

$∣ i ⟩$ , $∣ f ⟩$ = asymptotic in/out states (free particles in the far past / future),
$\hat{U} \to \hat{S}$ = the S-matrix, the unitary that maps in to out under the full interacting Hamiltonian.

The Born rule then reads $P_{i \to f} = ∣ ⟨ f ∣ \hat{S} ∣ i ⟩ ∣^{2}$ . No new postulate; just a specialization to asymptotic-state amplitudes.

Why the formula looks different

The QFT cross-section formula $d σ = (1/ F) \overline{∣ M ∣^{2}} d Π_{n}$ is the Born rule wrapped in three layers of bookkeeping:

Subtract the identity: scattering is $\hat{S} - 1$ , so $∣ ⟨ f ∣ \hat{T} ∣ i ⟩ ∣^{2} \propto ∣ (2 π)^{4} δ^{4} (\sum p) ∣^{2} \cdot ∣ M_{f i} ∣^{2}$ .
The squared delta function gives spacetime volume $V T$ . Dividing by $T$ converts probability to rate.
Dividing by the flux $F$ converts rate to cross section. Integrating over the final-state phase space sums over indistinguishable outcomes.

The cross-to-classical-rate "bridge" derivation in cross-sections.md §1.1 (iv) makes each step explicit.

Comparison with the QM Born rule and its cousins

The Born rule shows up in five recognizable forms across QM and QFT, all the same postulate with different bookkeeping:

Setting	Form
QM Postulate 3 (general)	$P (a) = ∥ ⟨ a ∥ ψ ⟩ ∥^{2}$
QM transition (Schrödinger picture)	$P_{i \to f} = ∥ ⟨ f ∥ \hat{U} (t, t_{0}) ∥ i ⟩ ∥^{2}$
Fermi's Golden Rule (NRQM, perturbative)	$Γ_{i \to f} = (2 π /ℏ) ∥ ⟨ f ∥ \hat{V} ∥ i ⟩ ∥^{2} ρ_{f} (E_{i})$
QFT scattering	$d σ = (1/ F) \overline{∥ M ∥^{2}} d Π_{n}$
QFT decay	$d Γ = (1/2 M) \overline{∥ M ∥^{2}} d Π_{n}$

Each of the last three has the structure (squared transition matrix element) × (sum over final states) × (kinematic conversion to a rate):

Element	Fermi's Golden Rule	QFT cross section
Squared matrix element	$∥ ⟨ f ∥ \hat{V} ∥ i ⟩ ∥^{2}$	$\overline{∥ M ∥^{2}}$
Density / measure of final states	$ρ_{f} (E_{i})$	$d Π_{n}$ (Lorentz-invariant phase space)
Conservation enforcer	implicit in $ρ_{f}$	$(2 π)^{4} δ^{4} (p_{f} - p_{i})$
Kinematic prefactor	$2 π /ℏ$	$1/ F$ (flux factor)

Fermi's Golden Rule is the non-relativistic single-particle scattering / decay limit of the QFT formulas. Both are the Born rule plus standard QM time-dependent perturbation theory; the QFT version adds Lorentz-covariant phase space, antiparticles, and Feynman-rule machinery for computing $M$ , but the underlying postulate is the same.

The Born rule appears in all the standard QM observables — transmission probabilities $T = ∥ t ∥^{2}$ , Rabi oscillation $P (t) = sin^{2} (Ω t /2)$ , decay rates, etc. The QFT cross section is just the most elaborate dressed-up version, decorated with relativistic and many-body bookkeeping.

A technical subtlety: non-normalizable states

The QM Born rule's literal probabilistic interpretation requires normalizable states for the inner product. In QFT:

Asymptotic in/out states are momentum eigenstates — not normalizable. $⟨ p ∣ p ⟩ = 2 E_{p} (2 π)^{3} δ^{3} (0)$ is a formal infinity (interpreted as proportional to the spatial volume $V$ ).
Squaring the S-matrix element produces $∣ δ^{4} ∣^{2}$ , also formally infinite (proportional to the spacetime volume $V T$ ).

The infinities cancel in the ratio (rate per particle pair) only because the delta-function squaring and the state-norm conventions pick up the same factors of $V$ and $T$ . This is why physicists work with probability per unit time per unit volume rather than probability directly; box-normalization (or any equivalent finite-spacetime regularization) is the rigorous way to derive the cross-section formula. Conceptually it is still the Born rule; mechanically, you need the regulator to interpret it.

In algebraic QFT (see § Algebraic QFT (Haag–Kastler)) the issue is sharper: local algebras are Type III, with no minimal projectors, so the literal $∣ ⟨ a ∣ ψ ⟩ ∣^{2}$ form of the Born rule does not even apply to local observables. One uses expectation values $ω (\hat{A}) = Tr (\overset{ρ}{^} \hat{A})$ directly, and cross sections are reconstructed from these.

Partial Reductions

Several frameworks make parts of the measurement axioms less fundamental, without fully eliminating them:

Decoherence. Tracing out the environment from a system + apparatus + environment composite produces an effectively diagonal density matrix in a "pointer basis" on a timescale $τ_{dec}$ . This explains why we never see macroscopic superpositions, but does not pick out a single outcome — that step still requires either the Born rule as an additional postulate, an interpretive move (Many-Worlds), or a dynamical-collapse model. QFT is the natural arena for decoherence, since the environment is typically a quantum field.
Gleason's theorem (1957). Any probability measure on the projection lattice of a Hilbert space (dimension $\geq 3$ ) satisfying positivity, normalization, and additivity over orthogonal projectors must take the form $P (\hat{Π}) = Tr (\overset{ρ}{^} \hat{Π})$ — the Born rule. So the form of the Born rule is forced once one accepts that probabilities exist; only their existence remains a postulate.
Many-Worlds derivations. Deutsch, Wallace, and Zurek (envariance) attempt to derive the Born rule within an Everettian framework with no collapse. Whether these arguments succeed is contested. They apply equally to QM and QFT.
Dynamical collapse models (GRW, CSL). Modify unitary evolution by adding a stochastic term that effects spontaneous collapse, replacing the collapse postulate with a dynamical equation. Lorentz-covariant relativistic versions are an active research area, but these are alternatives to QFT, not consequences of it.

Why It's Harder, Not Easier, in QFT

QFT actually complicates the orthodox measurement story in several ways:

Type-III local algebras. Local algebras $A (O)$ in algebraic QFT (see above) are typically Type III in the von Neumann classification. Type-III algebras have no minimal projectors and admit no decomposition of the identity into orthogonal one-dimensional projectors. So the collapse postulate as stated in QM — "project onto the eigenspace of $\hat{A}$ " — does not even quite apply to local observables in QFT.
Reeh–Schlieder theorem. In a relativistic QFT, every state can be approximated arbitrarily well by acting on the vacuum with operators localized in any bounded spacetime region. This means there is no clean tensor-factorization of the Hilbert space into "system" and "environment" — yet such a factorization is implicit in every textbook discussion of measurement.
Lorentz-covariant collapse is awkward. Naïve "instantaneous collapse on a spacelike hypersurface" picks a frame and breaks Lorentz invariance. Constructing a fully covariant collapse rule (or covariant CSL model) is a long-standing open problem.

Summary

The Born rule and collapse postulate sit in the same uncomfortable position in both QM and QFT: operationally indispensable, foundationally unexplained, and not derivable from the other postulates. QFT does not solve the measurement problem — it carries it forward, with extra technical complications from Type-III algebras and Lorentz covariance. Discussion of "measurement in QFT" is therefore really a discussion of how to apply QM-style measurement to QFT states, not a derivation of measurement from QFT first principles.

Keyboard shortcuts

youyuanwu