Skip to content

States, Observables, and Spectra

A quantum field is useful because its matrix elements probe states, not because every field component is directly observable. In a vacuum two-point function, inserting a complete set of energy–momentum states turns those matrix elements into a spectral measure. The measure then separates isolated stable-particle weight from multiparticle continua, and the Källén–Lehmann transform turns that separation into a precise distinction among poles, thresholds, and cuts.

This chapter follows the chain

matrix elementsspectral measureKa¨lleˊn–Lehmann formanalytic structure,\text{matrix elements} \longrightarrow \text{spectral measure} \longrightarrow \text{Källén–Lehmann form} \longrightarrow \text{analytic structure},

then asks where particle language succeeds or fails. A stable isolated mass shell, a resonance pole reached by analytic continuation, and an infraparticle threshold are three different structures. Only the first can enter the ordinary scattering route, and even then a pole is not by itself an LSZ theorem.

Choose an entry route · Read the spectral map · Review the chapter

Let Ω|\Omega\rangle be a translation-invariant vacuum and let O\mathcal O be a Hermitian scalar operator in a positive-metric physical Hilbert space. Subtract its vacuum expectation,

O^=OΩOΩ1,\widehat{\mathcal O} =\mathcal O -\langle\Omega|\mathcal O|\Omega\rangle\mathbf 1,

so the vacuum itself does not appear as a zero-momentum contribution. Translation covariance and a complete set of physical energy–momentum states resolve the Wightman function

WO(x)=ΩO^(x)O^(0)ΩW_{\mathcal O}(x) =\langle\Omega| \widehat{\mathcal O}(x)\widehat{\mathcal O}(0) |\Omega\rangle

into positive squared overlaps. With the spectrum in the closed forward cone and scalar Poincaré covariance, those overlaps assemble into a nonnegative measure ρO(μ2)dμ2\rho_{\mathcal O}(\mu^2)\,\mathrm d\mu^2. Whenever the time-ordered extension requires no additional local terms, its scalar Källén–Lehmann form is

D~F,O(p)=0dμ2ρO(μ2)ip2μ2+i0,ρO0.\widetilde D_{F,\mathcal O}(p) = \int_0^\infty \mathrm d\mu^2\, \rho_{\mathcal O}(\mu^2) \frac{i}{p^2-\mu^2+i0}, \qquad \rho_{\mathcal O}\ge 0.

The positivity statement uses the physical positive inner product and Hermiticity; translation covariance fixes the momentum phases; the spectrum condition fixes support; and scalar covariance reduces the measure to invariant mass. The completeness derivation and positive scalar measure are developed in Schwartz 2014, § 24.2.1, pp. 467–469. Distributional regularity is also required. Locality is not used in this basic two-point completeness argument, although it becomes essential in stronger QFT and scattering results. General composite time-ordered products can require subtractions or local contact terms; in that case the displayed equality is understood modulo those local terms, while the Wightman spectral measure remains the clean primary statement.

If O^\widehat{\mathcal O} has nonzero overlap with a stable scalar state of mass mm and that state is isolated from the continuum in the same quantum-number channel, then, in the simplest one-particle-plus-continuum case,

ρO(μ2)=ZOδ(μ2m2)+ρcont(μ2),ZO>0,suppρcont[s0,),s0>m2,\begin{aligned} \rho_{\mathcal O}(\mu^2) &=Z_{\mathcal O}\,\delta(\mu^2-m^2) +\rho_{\mathrm{cont}}(\mu^2), \\ Z_{\mathcal O}&>0, \qquad \operatorname{supp}\rho_{\mathrm{cont}} \subseteq [s_0,\infty), \qquad s_0>m^2, \end{aligned}

where s0s_0 is the lightest allowed multiparticle invariant mass in that channel. It need not be 4m24m^2. The transform becomes

D~F,O(p)=iZOp2m2+i0+is0dμ2ρcont(μ2)p2μ2+i0.\widetilde D_{F,\mathcal O}(p) =\frac{iZ_{\mathcal O}}{p^2-m^2+i0} +i\int_{s_0}^{\infty}\mathrm d\mu^2\, \frac{\rho_{\mathrm{cont}}(\mu^2)} {p^2-\mu^2+i0}.

The delta atom produces an isolated physical-sheet pole; continuous support produces a threshold branch point and cut in the infinite-volume limit. These structures may coexist. The residue ZOZ_{\mathcal O} depends on how O\mathcal O is normalized: rescaling the operator rescales both ZOZ_{\mathcal O} and ρO\rho_{\mathcal O}. Therefore neither ρO=1\int\rho_{\mathcal O}=1 nor ZO1Z_{\mathcal O}\le1 is asserted for an arbitrary operator. Schwartz 2014, § 24.2.1, p. 469; § 24.3, pp. 471–474 gives the pole-plus-continuum argument, while Weinberg 1995, § 10.2, pp. 430–434 explains why the pole criterion depends on a nonzero matrix element rather than on whether a field is called elementary or composite.

The chapter stops at three boundaries. It identifies a resonance with a complex pole on an analytically continued, channel-dependent nonphysical sheet—not with a normalizable unstable ket or merely a bump. It introduces an infraparticle as a long-range infrared case in which continuous spectral weight reaches the nominal mass and the isolated mass-shell atom can disappear—not as a resonance with a width. Finally, it states the stable-pole, overlap, asymptotic-state, separation, and infrared conditions that permit a scattering handoff, but it does not derive LSZ reduction. Detailed sheet analysis, infrared constructions, and scattering amplitudes continue in later volumes.

The overview imposes no prerequisite; each leaf states its own required and helpful background. Use the repair links below when a step is not yet routine.

Try thisReady whenRepair if unsure
Compare a field, an observable, and an interpolating operatorYou can explain why a field may have useful matrix elements without itself being a directly measurable observableQuantum states and operators repair, then Fields, Observables, and Interpolating Operators
Insert a complete set between two operatorsYou can identify which assumptions turn the terms into nonnegative squared overlaps and which operator quantum numbers restrict the intermediate statesVacua, States, and Representations
Normalize a one-particle state covariantlyYou can use $\langle\mathbf p’\mathbf p\rangle=2E_{\mathbf p}(2\pi)^3\delta^{(3)}(\mathbf p’-\mathbf p)$ without losing a phase-space factor
Compare Wightman and time-ordered two-point functionsYou can identify the ordering, vacuum, Fourier convention, and +i0+i0 prescription before interpreting a denominatorThe Generating Functional and Scalar Propagators, Ordered Correlators, and Sources
Read a positive spectral measure and a resolvent boundary valueYou can distinguish a point mass from continuous support and a physical boundary value from analytic continuation to another sheetSpectra, Resolvents, Spectral Measures, and Functional Calculus and Branches, Sheets, Analytic Continuation, and Monodromy

The six leaves have a coherent reading order, but the chapter has three useful entrances. “Required” below names an actual dependency; other sequencing is a recommendation chosen for the stated goal.

Reader goalRoute through the chapterObservable exit
Understand what operator data meanOperators, Observables, and Matrix ElementsSpectral Decomposition of Two-Point FunctionsThe Källén–Lehmann Representation. The first page requires the earlier states and interpolating-operator pages; the second additionally requires the generating functional.Insert completeness, derive the scalar spectral support, and state exactly why the measure is nonnegative
Translate spectral support into analytic structureSpectral DecompositionKällén–LehmannPoles, Cuts, Thresholds, and Stable Particles. Helpful mathematics comes from spectral measures, branches, and dispersion integrals.Map a delta atom to a stable pole and continuum support to a threshold cut without calling the cut a particle
Test particle language and scattering readinessPoles, Cuts, Thresholds, and Stable ParticlesResonances, Infraparticles, and Limits of Particle Language for the contrast. For the ordinary scattering route, combine the poles-and-cuts page with Fields, Observables, and Interpolating Operators, then read From One-Particle Poles to the Scattering Handoff.Decide whether the data support an ordinary stable external state, a resonance analysis, an infrared-modified framework, or no particle claim at all

Read the upper path in the figure from matrix elements to the Källén–Lehmann transform. Then compare the spectral components below it. An isolated delta atom and continuum support can occur together; the later resonance and infraparticle labels require additional analytic or infrared information and do not follow from the mere existence of a continuum.

Vacuum matrix elements and completeness produce a positive scalar spectral measure and its Källén–Lehmann transform; an isolated delta gives a stable physical-sheet pole that can reach LSZ only after extra asymptotic and infrared checks, while continuum support gives a threshold cut and may require separate resonance or infraparticle analysis.

Schematic spectral-content map for a vacuum-subtracted Hermitian scalar operator in a positive-metric physical Hilbert space. The stable delta atom and continuum may coexist. Resonance and infraparticle branches are conditional diagnoses, while the ordinary scattering handoff additionally requires an isolated shell, nonzero overlap, suitable in/out limits, and infrared control; the diagram is not a quantitative spectrum and is not to scale.

The same relationships can be read without the image.

Input or featureLicensed conclusionWhat does not follow
Vacuum matrix elements, completeness, translation covariance, the forward spectrum, scalar covariance, and a positive physical inner productA nonnegative invariant-mass measure for a Hermitian scalar operatorLocality, clustering, a mass gap, or asymptotic completeness
Källén–Lehmann transform of that measureA time-ordered two-point function built from free-mass denominators with the common +i0+i0 boundary valuePositivity for gauge-variant components in an indefinite auxiliary space or for arbitrary off-diagonal and non-vacuum correlators
Isolated ZOδ(μ2m2)Z_{\mathcal O}\delta(\mu^2-m^2)A simple physical-sheet pole with nonnegative operator-dependent residue and a stable one-particle contributionThat the state is elementary, that ZO1Z_{\mathcal O}\le1, or that LSZ assumptions are already satisfied
Continuous support from s0s_0A threshold branch point and cut in the infinite-volume limitA new particle at the threshold, or automatically a resonance or infraparticle
Pole reached by continuation through a channel cutA resonance diagnosis with explicit sheet and channel conventionsA normalizable unstable one-particle ket or an ordinary LSZ external state
Continuous threshold behavior at a nominal charged mass because arbitrarily soft quanta cannot be separatedPossible infraparticle behavior with no isolated mass-shell poleA decay width, a resonance interpretation, or one universal threshold exponent
Stable isolated shell, finite nonzero overlap, suitable asymptotic limits and separation, plus infrared and physical-state controlReadiness to begin the ordinary scalar scattering reductionThe LSZ derivation itself or asymptotic completeness of the whole theory

The guide below lists every leaf once in its dependency order.

  1. Operators, Observables, and Matrix Elements. Required background is Vacua, States, and Representations together with Fields, Observables, and Interpolating Operators. Distinguish an operator from a measurement prescription, identify the state dependence of matrix elements, and explain how an interpolating operator can reveal a particle sector without being an observable in every sense.

  2. Spectral Decomposition of Two-Point Functions. Required background is the first page and The Generating Functional. Helpful background comes from Bounded, Compact, and Integral Operators and Spectra, Resolvents, Spectral Measures, and Functional Calculus. Insert completeness, use translation covariance, and derive the positive scalar spectral sum or measure before introducing the propagator representation.

  3. The Källén–Lehmann Representation. Required background is the spectral-decomposition page; spectra and resolvents are helpful. Derive the scalar representation with its minimal vacuum, covariance, spectrum, completeness, positivity, and regularity assumptions visible. Identify which steps do not use locality and where composite, gauge-fixed, or non-vacuum cases require new qualifications.

  4. Poles, Cuts, Thresholds, and Stable Particles. Required background is Källén–Lehmann. Helpful background comes from Spectra, Resolvents, Spectral Measures, and Functional Calculus, Branches, Sheets, Analytic Continuation, and Monodromy, and Boundary Values, Discontinuities, and Dispersion Integrals. Translate discrete and continuous spectral support into the physical-sheet pole, threshold, and cut dictionary, including the finite-volume qualification.

  5. Resonances, Infraparticles, and Limits of Particle Language. Required background is the poles-and-cuts page. Contrast an analytically continued resonance pole with an infrared threshold lacking an isolated mass shell. This page supplies the conceptual warning; detailed resonance extraction continues in Scattering, and developed infraparticle theory continues in Scattering, Gauge Theories, and Mathematical QFT.

  6. From One-Particle Poles to the Scattering Handoff. Required background is the poles-and-cuts page together with Fields, Observables, and Interpolating Operators. State the pole, residue, normalization, asymptotic-state, isolation, and infrared hypotheses needed before ordinary scattering reduction can begin. Resonances and infraparticles are failure or modification cases, but page 5 is not a hard prerequisite for this stable-particle route.

The chapter inherits the site’s (+)(+---) metric. The following page-local data prevent the same spectral symbol from silently changing meaning.

DatumChapter conventionCheck before reuse
Vacuum subtraction$\widehat{\mathcal O}=\mathcal O-\langle\Omega\mathcal O
State normalization$\langle\mathbf p’\mathbf p\rangle=2E_{\mathbf p}(2\pi)^3\delta^{(3)}(\mathbf p’-\mathbf p)$
Fourier transformf(p)=d4xeipxf(x)f(p)=\int\mathrm d^4x\,e^{ip\cdot x}f(x), with inverse phase eipxe^{-ip\cdot x}A source using the opposite phase has all momentum labels translated consistently
Ordering and state$D_{F,\mathcal O}=\langle\OmegaT\widehat{\mathcal O}\widehat{\mathcal O}
Spectral variableThe measure is written ρO(μ2)dμ2\rho_{\mathcal O}(\mu^2)\,\mathrm d\mu^2 on μ20\mu^2\ge0Factors associated with a density in μ\mu rather than μ2\mu^2 are not dropped
PositivityρO0\rho_{\mathcal O}\ge0 for a Hermitian scalar operator in the declared positive physical spaceGauge-variant, ghost, non-Hermitian, off-diagonal, and non-vacuum cases are separately qualified
Boundary valueEvery scalar denominator is i/(p2μ2+i0)i/(p^2-\mu^2+i0)A physical boundary value is distinguished from continuation through a cut
Residue symbolZOZ_{\mathcal O} is the operator’s one-particle overlap weightIt is not the source functional Z[J]Z[J] and is not normalization-independent
Sheet language“Physical sheet” refers to the boundary value fixed by +i0+i0; another sheet is reached by a declared analytic continuationA resonance pole is never drawn or described as a first-sheet stable pole

For the canonically normalized free real scalar, take O=ϕ\mathcal O=\phi and ΩϕΩ=0\langle\Omega|\phi|\Omega\rangle=0. Completeness has one one-particle contribution in the field’s channel and no interacting multiparticle continuum, so

ρ0(μ2)=δ(μ2m2),D~F(0)(p)=ip2m2+i0.\rho_0(\mu^2)=\delta(\mu^2-m^2), \qquad \widetilde D_F^{(0)}(p) =\frac{i}{p^2-m^2+i0}.

This is the round-trip check for every later convention: the spectral integral must reproduce the same Feynman function obtained canonically and from the generating functional. If the field is rescaled, ϕcϕ\phi\mapsto c\phi, the density becomes c2δ(μ2m2)c^2\delta(\mu^2-m^2); positivity survives, but the normalization changes.

An interacting operator can retain an isolated atom while acquiring continuum weight,

δ(μ2m2)ZOδ(μ2m2)+ρcont(μ2).\delta(\mu^2-m^2) \quad\longrightarrow\quad Z_{\mathcal O}\delta(\mu^2-m^2) +\rho_{\mathrm{cont}}(\mu^2).

That single change supplies the chapter’s recurring comparison. The atom gives the stable pole. The continuum gives the threshold cut. A resonance requires analytic continuation beyond that physical boundary, and an infraparticle instead removes the isolated sharp-mass contribution through long-range infrared structure. None of those last three statements is obtained by renaming the free pole.

QuestionStrongest conclusion in this chapterCategory error prevented
What physical information does a field carry?Its state-dependent matrix elements can probe sectors with matching quantum numbersEvery field component is an observable
Why is the scalar spectral measure positive?Its weights are squared overlaps in a positive physical Hilbert space for a Hermitian scalar operatorEvery propagator or spectral matrix is entrywise positive
What does an isolated spectral atom mean?It gives a stable one-particle contribution and a physical-sheet pole when the mass shell is isolatedEvery pole is a particle, regardless of sheet or state status
What does a continuum mean?In infinite volume it produces threshold support and a cutEvery threshold or cut is a particle
What distinguishes a resonance?A channel- and sheet-qualified pole appears after analytic continuation; no normalizable unstable one-particle ket is impliedA line-shape bump is automatically a state or a resonance pole
What distinguishes an infraparticle?Long-range soft structure can replace the isolated mass shell by continuous threshold behaviorAn infraparticle is merely a resonance with a small width
When may ordinary scattering reduction begin?Only after stable-pole, overlap, asymptotic-state, isolation, wave-packet, infrared, and physical-state conditions are statedThe handoff is already the LSZ derivation or a proof of asymptotic completeness

For resonances, the relevant distinction between physical and unphysical sheets—and the warning that a bump is neither necessary nor sufficient—is summarized in Particle Data Group 2025, “Resonances,” § 50.1.1, pp. 3–6 (PDF). For the infrared contrast, Buchholz 1986, pp. 331–334 proves, under its Gauss-law hypotheses, that charged states cannot be mass-operator eigenstates. The chapter uses that result as a bounded example, not as a universal claim about every massless theory.

Use the stated criteria to identify exactly which physical or analytic step needs revision, then follow the corresponding repair route.

CheckA successful response must showRepair route
Operator classificationGive one field that serves as an interpolating operator, state which matrix element matters, and explain why this alone does not define a measurementFields, Observables, and Interpolating Operators
Positivity derivationInsert completeness into WO(x)W_{\mathcal O}(x) and point to the exact use of Hermiticity, the positive inner product, translation covariance, and the spectrum conditionSpectral Decomposition of Two-Point Functions
Free normalizationSubstitute ρ0(μ2)=δ(μ2m2)\rho_0(\mu^2)=\delta(\mu^2-m^2) into the Källén–Lehmann integral and recover i/(p2m2+i0)i/(p^2-m^2+i0)The Källén–Lehmann Representation
Singularity classificationGiven an isolated delta plus continuum from s0s_0, identify the pole and cut, and explain why the threshold itself is not a particlePoles, Cuts, Thresholds, and Stable Particles
Failure diagnosisContrast a second-sheet resonance pole with continuous spectral behavior beginning at a nominal charged mass; state why neither is an ordinary stable external ketResonances, Infraparticles, and Limits of Particle Language
Scattering readinessList the data missing from the statement “the two-point function has a pole” before an ordinary LSZ calculation may beginFrom One-Particle Poles to the Scattering Handoff
Computational designFor a supplied discrete-plus-continuum measure, specify a quadrature, convergence test, error estimate, and the continuum claim that finite data would not establishSpectral Decomposition of Two-Point Functions
  • Buchholz, Detlev. “Gauss’ Law and the Infraparticle Problem.” Physics Letters B 174 (1986): 331–334. DOI.

  • Particle Data Group. “Resonances.” In Review of Particle Physics, 2025 update. Official PDF.

  • Schwartz, Matthew D. Quantum Field Theory and the Standard Model. Cambridge: Cambridge University Press, 2014. DOI.

  • Weinberg, Steven. The Quantum Theory of Fields, Volume I: Foundations. Cambridge: Cambridge University Press, 1995. DOI.