Asymptotic Completeness in Rayleigh Scattering

• The paper establishes asymptotic completeness by proving that below the ionization threshold, every state evolves into a bound particle state accompanied by a free photon cloud.
• It employs rigorous propagation estimates, minimal escape velocity bounds, and the Deift–Simon wave operator method to control infrared singularities.
• These results have critical implications for spectral analysis and advance our understanding of photon emission control in quantum field models.

Asymptotic completeness of Rayleigh scattering is the statement that, for non-relativistic particle–field systems at energies below the ionization threshold, the full long-time dynamics is exhausted by channels in which the matter subsystem remains bound while the radiation field becomes asymptotically free. In non-relativistic QED and Nelson-type models, this is the rigorous version of the picture in which an excited atom or molecule relaxes to a dressed bound state and the excess energy is carried away by escaping massless quanta. The subject is technically dominated by the infrared problem: one must control the long-time production of soft photons strongly enough to construct wave operators and to prove that no additional below-threshold channels remain (Faupin et al., 2012, Faupin et al., 2012, Griesemer et al., 24 Dec 2025).

1. Definition and below-threshold scattering picture

In the Rayleigh-scattering regime, the Hamiltonian acts on a tensor product of a particle Hilbert space and a bosonic Fock space, and the relevant spectral subspace is the range of the spectral projection below the ionization threshold. Physically, this is the regime in which the particles remain bound, while the field may contain asymptotically free photons. Mathematically, asymptotic completeness means that the scattering states produced by the dynamics exhaust the physically relevant below-threshold subspace, or equivalently that every admissible asymptotic configuration consisting of a bound subsystem plus free radiation is realized by an actual solution (Faupin et al., 2012).

The early Rayleigh-scattering formulation is explicit. For states below threshold, the Schrödinger evolution is approximated asymptotically by vectors of the form

$e^{-it\mathcal E_j}\,\phi_j \otimes e^{-itH_f}f_j,$

with finitely many photons in $f_j$, and the approximation error tends to zero in the sense stated in the completeness theorem. In this form, asymptotic completeness says that every state in $\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$ asymptotically becomes a superposition of bound matter states with an outgoing photon cloud (Faupin et al., 2012).

A recurrent source of confusion is the scope of the theorem. The relevant completeness statement is not full many-body scattering at arbitrary energy. It is explicitly restricted to energies below the ionization threshold, where the matter sector stays bound and only the radiation field scatters. This is why the result is called asymptotic completeness for Rayleigh scattering rather than a general ionization or Compton-type scattering theorem (Faupin et al., 2012).

2. Model classes and spectral hypotheses

The standard non-relativistic QED Pauli–Fierz Hamiltonian considered in the detailed photon case has the form

$$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$

acting on $\mathcal H=\mathcal H_p\otimes\mathcal F$, with $\mathcal F=\Gamma_s(\mathfrak h)$ and one-photon space $\mathfrak h=L^2(\mathbb R^3;\mathbb C^2)$. The assumptions include self-adjointness of the particle Hamiltonian, existence of a unique ground state, and a spectral threshold $E$ below which the particle subsystem remains bound and exponentially localized (Faupin et al., 2012).

A closely related massless Nelson model is formulated on

$$\mathcal{H} = L^2(\mathbb{R}^3,dx)\otimes \mathcal{F}(h),\qquad h=L^2(\mathbb{R}^3,dk),$$

with Hamiltonian

$$H = H_p \otimes 1 + 1\otimes H_\omega + g\,\phi(w_x),$$

where $f_j$0, $f_j$1, $f_j$2, and $f_j$3. The main theorem in that setting assumes a simple ground state, a positive gap $f_j$4 to the ionization threshold for small coupling, and $f_j$5, a condition stronger than the ground-state requirement $f_j$6 and used to obtain the propagation estimates needed for the scattering argument (Griesemer et al., 24 Dec 2025).

The general framework of the earlier Rayleigh-scattering theorem is broader. It treats Hamiltonians

$f_j$7

with linear coupling

$f_j$8

under an infrared/regularity hypothesis on $f_j$9, a localization estimate below the ionization threshold, a Fermi Golden Rule condition excluding embedded eigenvalues in $\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$0, and sufficiently small coupling. The massless case $\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$1 is the principal difficulty; phonon variants and the spin-boson model are included as examples (Faupin et al., 2012).

3. Propagation estimates, Huygens-type bounds, and wave operators

The decisive mechanism in the proof of Rayleigh-scattering completeness is propagation away from the matter subsystem. The central estimates show that photons which are not part of the bound cloud cannot linger near the particles for large times. In the original framework this appears as a weak minimal escape estimate of the form

$\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$2

together with integrated “quantum Huygens principle” bounds such as

$\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$3

These estimates express that subluminal photons do not contribute asymptotically near the interaction region (Faupin et al., 2012).

In the Pauli–Fierz photon case, the minimal photon velocity bound and the weak minimal escape velocity estimate are combined with maximal velocity bounds from earlier work. The resulting picture is that escaping photons neither remain trapped near the matter subsystem nor outrun the light cone. This is precisely the dynamical separation required by the Deift–Simon construction: one can partition the field into near and far components and identify the far component with the free asymptotic radiation (Faupin et al., 2012).

The wave operator is built on an enlarged space by a second-quantized partition of unity. In the original Rayleigh-scattering paper it is

$\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$4

while later below-threshold QED formulations use

$\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$5

The proof reduces to showing that the family $\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$6 is Cauchy. The derivative splits into a main propagation term, controlled by minimal-velocity estimates, and an interaction remainder, whose treatment requires quantitative photon-number information (Faupin et al., 2012, Faupin et al., 2012).

A major refinement in the massless Nelson model is the use of the Jakšić–Pillet expansion trick. One passes to an enlarged Fock space containing “fake bosons” of negative energy, with

$\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$7

and rewrites the field Hamiltonian as $\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$8. In these variables, localization in the radial coordinate $\operatorname{Ran}E_{(-\infty,\mathcal E)}(H)$9 and commutator estimates become substantially simpler. The new propagation estimate

$$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$0

is the key technical input that makes a weaker photon-number hypothesis sufficient for constructing the Deift–Simon wave operator (Griesemer et al., 24 Dec 2025).

4. Photon-number control and the shift from expectation bounds to tail criteria

The central obstruction in massless models is lack of sufficient control on the number of emitted soft photons. Early completeness theorems assumed a uniform photon-number bound, for example

$$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$1

This assumption is natural, but it is stronger than the dynamics actually uses in the Deift–Simon argument (Faupin et al., 2012, Faupin et al., 2012).

Criterion	Representative form	Role
Uniform expectation bound	$$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$2	Early sufficient hypothesis
Localized or dense-set bound	Condition (i) or (i′) for spectrally localized states or a dense set	Weakened sufficient hypothesis
Photon-tail tightness	$$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$3 as $$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$4	State-wise necessary-and-sufficient criterion

A technically important correction weakened the original global assumption. The proof of asymptotic completeness below the ionization threshold still works if the photon-number bound is assumed only for states spectrally localized strictly below the threshold, or even on a dense set $$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$5 with a constant depending on the initial vector. The commutator analysis is then rebuilt around a weighted low-momentum estimate,

$$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$6

and a weighted observable such as

$$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$7

which is better adapted to the infrared behavior of massless bosons. For a class of Hamiltonians, including the massless spin-boson model, these weaker bounds can be verified by adapting the De Roeck–Kupiainen strategy (Faupin et al., 2012).

The sharpest formulation appears in the massless Nelson model. There the main theorem identifies the precise criterion for a given state $$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$8: $$H=\sum_{j=1}^n \frac1{2m_j}\bigl(-i\nabla_{x_j}-k_j A(x_j)\bigr)^2+U(x)+H_f, \qquad H_f=d\Gamma(\omega),\quad \omega(k)=|k|,$$9 This condition controls the tails of the photon-number distribution rather than the expectation of $\mathcal H=\mathcal H_p\otimes\mathcal F$0. It is therefore strictly weaker than a uniform bound on $\mathcal H=\mathcal H_p\otimes\mathcal F$1, and it is also necessary, not merely sufficient. A plausible implication is that the correct infrared object for Rayleigh scattering is tightness of the photon-number distribution at large times, rather than uniform boundedness of its first moment (Griesemer et al., 24 Dec 2025).

5. Established theorems and model-dependent variants

The original completeness theorem proves that, under the infrared hypotheses, the localization estimate below threshold, the Fermi Golden Rule condition, small coupling, and the uniform photon-number bound, asymptotic completeness holds on

$\mathcal H=\mathcal H_p\otimes\mathcal F$2

The asymptotic state is a bound particle state accompanied by finitely many outgoing photons, and the proof proceeds through propagation observables, minimal escape estimates, and the Deift–Simon wave operator (Faupin et al., 2012).

The photon case for the standard Pauli–Fierz Hamiltonian requires an additional infrared normalization step. The detailed QED treatment uses the generalized Pauli–Fierz transform

$\mathcal H=\mathcal H_p\otimes\mathcal F$3

to conjugate the Hamiltonian to a form with improved infrared behavior, after which the Rayleigh-scattering proof follows the phonon/Nelson strategy. This fills in the proof details for standard non-relativistic QED below the ionization threshold, still under small coupling, the Fermi Golden Rule, and a uniform bound either on $\mathcal H=\mathcal H_p\otimes\mathcal F$4 or on $\mathcal H=\mathcal H_p\otimes\mathcal F$5 on a dense set (Faupin et al., 2012).

The massless Nelson result strengthens the logical structure. Assuming $\mathcal H=\mathcal H_p\otimes\mathcal F$6, $\mathcal H=\mathcal H_p\otimes\mathcal F$7, small $\mathcal H=\mathcal H_p\otimes\mathcal F$8, and the vacuum decay or Mourre-type input

$\mathcal H=\mathcal H_p\otimes\mathcal F$9

the theorem identifies scattering states exactly by the photon-tail criterion. The proof also establishes a minimal escape property,

$\mathcal F=\Gamma_s(\mathfrak h)$0

showing that photons cannot remain too close to the electron for too long (Griesemer et al., 24 Dec 2025).

A structurally simpler but conceptually related result holds for translation-invariant Nelson-type and Polaron-type models restricted to the vacuum and one-particle sectors. There one proves asymptotic completeness with

$\mathcal F=\Gamma_s(\mathfrak h)$1

and

$\mathcal F=\Gamma_s(\mathfrak h)$2

The proof is fiberwise in the total momentum $\mathcal F=\Gamma_s(\mathfrak h)$3, using an HVZ-type theorem, a threshold set

$\mathcal F=\Gamma_s(\mathfrak h)$4

strict Mourre estimates away from thresholds, and geometric asymptotic completeness. This model is explicitly described as Rayleigh-scattering-like because the only asymptotic channel is a bound dressed particle plus one free boson (Gérard et al., 2015).

6. Scope, unresolved difficulties, and distinct usages of the term

The principal unresolved issue in full massless Rayleigh scattering is still infrared control of emitted soft photons. One formulation states directly that asymptotic completeness of Rayleigh scattering in models of atoms and molecules of non-relativistic QED is expected, but that a proof lacks sufficient control on the number of emitted soft photons except in the spin-boson model. Earlier results therefore remain conditional on photon-number bounds, and later work shows how to weaken or reinterpret those conditions rather than eliminate infrared analysis altogether (Griesemer et al., 24 Dec 2025, Faupin et al., 2012).

Another important limitation is that the established theorems concern the below-threshold sector. They do not describe ionization channels, asymptotic free charged particles, or the full many-body scattering problem above threshold. In this sense, asymptotic completeness for Rayleigh scattering is a sharply delimited theorem about bound matter plus outgoing radiation, not a universal completeness statement for all spectral regimes (Faupin et al., 2012).

The terminology also has a distinct, non-wave-operator usage in spectroscopic studies of polarized radiation. In that setting, “asymptotic completeness” refers to completeness of a spectral series in the far wings: if a group of transitions is complete in the dipole sense and upper-term interference is included, then the far-wing redistribution becomes Rayleigh-like and the linear polarization approaches the asymptotic limit

$\mathcal F=\Gamma_s(\mathfrak h)$5

for $\mathcal F=\Gamma_s(\mathfrak h)$6 scattering of a collimated, unpolarized beam. This is an application of spectroscopic stability, not a statement about the range of scattering wave operators. The shared phrase thus names two different asymptotic closure phenomena: one dynamical and operator-theoretic, the other spectroscopic and angular-momentum-theoretic (Casini et al., 2017).