Clustering, Vacuum Uniqueness, and Mass-Gap Implications

Clustering controls the vacuum sector, while the rate of Euclidean-time clustering probes the spectrum of the reconstructed Hamiltonian. Qualitative clustering can imply uniqueness of the translation-invariant vacuum under cyclicity and density hypotheses; a uniform exponential bound can imply a positive excitation gap. Neither statement alone produces an isolated one-particle mass shell.

Required background. Positivity, spectrum, covariance, and locality hypotheses supplies spectral support and positivity. Osterwalder–Schrader reconstruction supplies the vacuum representation and $H\geq0$. Euclidean growth, regularity, and temperedness supplies the bounds needed to pass from Schwinger functions to operators and distributions.

Helpful background. Clustering, vacuum assumptions, and long-range correlations gives the physical distinctions. The nonabelian mass gap as a spectral statement explains why a gap claim must identify its Hilbert space and observable sector.

Qualitative clustering and vacuum uniqueness

For positive-time observables $A,B$, reflection reconstruction gives connected Euclidean-time correlations of the form

$$C_{A,B}(\tau) =\langle A\Omega,e^{-\tau H}(1-P_0)B\Omega\rangle, \qquad \tau>0,$$

where $P_0$ projects onto $\ker H$. A qualitative cluster property requires, for a declared translation direction and suitable observables,

$$\lim_{\tau\to\infty} \left(\langle A\,T_E(\tau)B\rangle_E -\langle A\rangle_E\langle B\rangle_E\right)=0.$$

If the reconstructed field algebra is cyclic on $\Omega$ and the cluster statement holds on a dense set, then the only invariant vector in the vacuum sector is proportional to $\Omega$: otherwise the projection onto another invariant vector would leave a nonfactorizing large-separation component. In measure language, clustering is associated with an extremal, or pure-phase, translation-invariant Euclidean state. A mixture of two phases can remain Euclidean invariant and reflection positive but fails clustering because the phase label is correlated at arbitrary distance.

This is why clustering should not be hidden inside the construction of the Hilbert space. The quotient and positive semigroup exist before it; clustering sharpens the statement about $P_0$. Osterwalder and Schrader 1973, §§4.4–4.5, pp. 96–97 give the reconstructed cluster and locality steps.

Exponential decay and a Hamiltonian gap

For $A=B^*$, the spectral theorem supplies a positive measure

$$d\mu_A(E)=d\langle(1-P_0)A\Omega,P_H(E)(1-P_0)A\Omega\rangle$$

such that

$$C_{A^*,A}(\tau)=\int_{(0,\infty)}e^{-\tau E}\,d\mu_A(E).$$

If $\operatorname{spec}(H)\cap(0,m)=\varnothing$, then

$$\lvert C_{A,B}(\tau)\rvert \leq e^{-m\tau} \lVert(1-P_0)A^*\Omega\rVert \lVert(1-P_0)B\Omega\rVert.$$

Conversely, suppose a bound with rate $m>0$ holds for positive diagonal correlations on a set whose vectors are dense in $\Omega^\perp$, with constants that do not degrade the exponential rate. Positivity of $d\mu_A$ then forbids spectral support below $m$: any mass in $(0,m-\varepsilon)$ would make $e^{m\tau}C_{A^*,A}(\tau)$ grow exponentially. Density removes that spectral interval for $H$ itself.

The quantifiers matter. Exponential decay for one observable excludes low energy only from that observable’s spectral measure; it can miss states orthogonal to the chosen channel. A polynomial prefactor does not change the threshold exponent, while a bound known only on a finite range of $\tau$ proves no exact gap. Spatial exponential decay can be converted to a relativistic mass statement only after full covariance and the relevant analytic/spectral representation are established.

First QFT application: the weakly coupled P(φ)₂ model

For a stable massive $P(\phi)_2$ interaction at sufficiently small coupling and in the selected pure phase, cluster expansions give exponential decay of truncated Schwinger functions. Applied to a dense family of local polynomial observables, the preceding Laplace-transform argument yields a one-dimensional vacuum sector and a positive lower bound on the reconstructed excitation spectrum.

Rigorous construction status and open problems supplies the broader dimensional and model-dependent boundary.

The model-specific result is stronger: Glimm, Jaffe, and Spencer proved the Wightman axioms and found isolated mass-operator eigenvalues at $0$ and at a positive mass $m$ in the weak-coupling regime Glimm, Jaffe, and Spencer 1974, pp. 585–632. That particle conclusion uses their additional spectral and cluster-expansion analysis; it is not a consequence of exponential clustering alone.

There is also a useful current boundary. A 2025 stochastic-quantization construction obtains infinite-volume planar $P(\Phi)_2$ accumulation points with Euclidean invariance, reflection positivity, and OS regularity for general bounded-below even-degree interactions, but explicitly does not prove clustering or uniqueness of the infinite-volume limit in that generality Duch, Dybalski, and Jahandideh 2025, Theorem 1.1 and Remarks 1.3–1.4. One must not transfer the weak-coupling pure-phase gap conclusion to every such accumulation point.

Adversarial test: decay without a particle pole

Consider a generalized free field with Källén–Lehmann measure

$$d\rho(\mu^2)=\mathbf 1_{[m^2,\infty)}(\mu^2),r(\mu^2)\,d\mu^2, \qquad m>0,$$

where $r\geq0$ is locally integrable and has suitable polynomial growth, but $d\rho$ has no atom. Its Euclidean two-point function is a positive superposition of massive free covariances and is reflection positive. The reconstructed spectrum begins at $m$, so correlations decay with exponential scale $e^{-m\tau}$, generally multiplied by a power of $\tau$ determined by threshold behavior.

There is nevertheless no delta-function contribution $Z\delta(\mu^2-M^2)$ and hence no isolated one-particle pole. The strongest conclusion licensed by the decay is a spectral threshold or gap in the tested sector. Establishing a particle requires an isolated mass hyperboloid, nonzero spectral projection, and the relevant covariance and locality hypotheses.

Independent checks

• Connected part: subtract the correct vacuum expectation or project with $1-P_0$ before fitting a decay rate.
• Channel coverage: show that the tested observables generate a dense subspace, or state that the gap conclusion is channel-specific.
• Uniform regime: distinguish an asymptotic theorem from finite-distance numerical behavior and track cutoff and volume dependence.
• Phase: verify extremality or select a pure phase; a convex mixture can violate clustering without violating reflection positivity.
• Particle claim: inspect whether the spectral measure contains an atom, not merely whether its support has a positive lower edge.

Exercise

Let $d\mu$ be a finite positive measure on $(0,\infty)$ and $C(\tau)=\int e^{-\tau E}d\mu(E)$. Show that $C(\tau)\leq Ke^{-m\tau}$ for all sufficiently large $\tau$ implies $\mu((0,m-\varepsilon))=0$ for every $\varepsilon>0$.

Solution

If $a=\mu((0,m-\varepsilon))>0$, then

$$C(\tau)\geq\int_{(0,m-\varepsilon)}e^{-\tau E}d\mu(E) \geq a e^{-(m-\varepsilon)\tau}.$$

Combining this with $C(\tau)\leq Ke^{-m\tau}$ gives $a e^{\varepsilon\tau}\leq K$, impossible as $\tau\to\infty$. Thus $a=0$ for every $\varepsilon>0$, and the measure has no support below $m$.

References

• Duch, Paweł, Wojciech Dybalski, and Azam Jahandideh. “Stochastic Quantization of Two-Dimensional $P(\Phi)$ Quantum Field Theory.” Annales Henri Poincaré 26 (2025): 1055–1086. doi:10.1007/s00023-024-01447-w. Open article.
• Glimm, James, Arthur Jaffe, and Thomas Spencer. “The Wightman Axioms and Particle Structure in the $\mathscr P(\phi)_2$ Quantum Field Model.” Annals of Mathematics 100 (1974): 585–632. doi:10.2307/1970959. Journal page.
• Osterwalder, Konrad, and Robert Schrader. “Axioms for Euclidean Green’s Functions.” Communications in Mathematical Physics 31 (1973): 83–112. doi:10.1007/BF01645738. Open PDF.