{"id":"732d9323-2268-45eb-812c-bf9592b1684f","arxiv_id":"2607.29485","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For Friedrichs Hamiltonians with discrete bath spectra converging weakly to a Hölder-continuous density, the survival amplitude is approximately the Fermi-golden-rule exponential with explicit error, and cavity decay appears when the coupling exceeds the square root of the spectral spacing.","lead":"This paper proves that a quantum state coupled to a bath with discrete but nearly dense energy levels can still decay almost exponentially on intermediate timescales, with the rate set by Fermi's golden rule. The result gives a rigorous threshold for when exponential decay appears in atoms inside large cavities, with explicit error bounds in terms of coupling strength and spectral denseness.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The uniform-in-time contour estimate in the β>1/2 proof of Theorem 4.1 is unjustified: e^{-itz} grows on the upper arc of γ(L) and real-axis poles inside the contour carry the t-dependence, so the no-decay threshold β=1/2 is not established.","rationale":"The reader's formal weakest_assumption focuses on the scale separation d≪μ² in Assumption (Aα), which is important for the regime where Theorem 1.1 gives nontrivial decay; that part of the proof appears internally consistent. The most load-bearing concern is instead the β>1/2 no-decay claim in Theorem 4.1, which the reader mentions in the rationale but does not list as the weakest assumption. The contour-integral step as written is a genuine gap: it bounds the integrand on a contour where |e^{-itz}| is exponentially large for the times that matter, and it ignores residues at real-axis poles. If this gap is real, the sharpness of the β<1/2 condition is unproven, and the paper's advertised threshold may be incorrect. The verdict should remain CONDITIONAL: the main theorem may survive, but the sharpness claim needs either a correct proof or a revision. The proposed numerical toy test would settle whether the uniform bound is actually false.","tokens_in":18629,"tokens_out":33079,"duration_ms":357305,"concrete_test":"Fix β=0.75 and a finite toy Friedrichs model: K_L on ℓ²({−L²,...,L²}) with eigenvalues E_j=E0+jL^{-2}, weights w_j=c(1+j²/L^4)^{-1} normalized so Σw_j=1, and μ_L=L^{-β}. Diagonalize the rank-one Friedrichs Hamiltonian H_L(μ_L) for L=50,100,200,400 and compute a_L(1)=⟨χ,e^{-i μ_L^{-2}(H_L−E0)}χ⟩. If a_L(1) converges to e^{-Γ(E0)} (or any nontrivial decay) instead of 1, the uniform-in-time estimate in Theorem 4.1 fails; if it converges to 1, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Theorem 4.1 for β∈(1/2,∞), the authors choose a contour γ(L) of radius ε_L∈(1/2,3/2) centered at E0, bound |Σ_L(z)|≤C2L on it, and then assert that the Neumann-series representation of ⟨χ,e^{-itH_L}P_Lχ⟩ gives |⟨χ,e^{-itH_L}P_Lχ⟩−e^{-itE0}|≤C4 L^{1−2β} uniformly in t. This does not follow. The contour has arcs with Im z>0, so |e^{-itz}|=e^{t Im z} grows like e^{O(1)t}; pointwise control of the integrand on γ(L) is t-independent and cannot yield a t-uniform bound. Moreover, Σ_L(z) has poles at σ(K_L) on the real axis inside γ(L), so the contour cannot be deformed to the lower half-plane without picking up residues e^{-itλ_j}. At the relevant rescaled times t=τ μ_L^{-2}=τ L^{2β}, phases across the resonance width vary by O(τ), so the amplitude should generically approach e^{-Γ(E0)τ}, not 1. The claimed sharp threshold β=1/2 is therefore unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Friedrichs Hamiltonians whose bath spectral measures are discrete but converge weakly to an absolutely continuous limit with Hölder density. Theorem 1.1 asserts that the survival amplitude of the excited state is approximately exponential at rate Γ(E0)=πρ∞(E0) and energy shift Δ(E0), with an error expressed through the coupling μ, the Hölder exponent α, and the Lévy distance d(ν_L,ν∞). Corollary 1.1 derives a simultaneous weak-coupling/large-volume (van Hove) limit for μ_L=d(ν_L,ν∞)^β, β<1/2. Section 4 applies this to the rotating-wave confined spin-boson model, with Lemma 4.1 supplying quantitative convergence of the spectral measures and Theorem 4.1 claiming exponential decay for β<1/2 but no decay (amplitude 1) for β>1/2.","tokens_in":18953,"tokens_out":21307,"duration_ms":201824,"significance":"If the main theorem is correct, it gives a rigorous finite-volume mechanism for the emergence of exponential decay before taking the thermodynamic limit, with explicit error estimates in terms of the Lévy distance and no fitted parameters. The proof strategy—smearing the spectral measure by a Cauchy kernel, then comparing the resulting density to a Lorentzian via self-energy regularity—is natural and largely self-contained. Lemma 4.1 provides an explicit O(L^{-1}) Lévy-distance bound for the spin-boson application, which is a useful quantitative result. Theorem 1.1 and the existence part of Corollary 1.1 appear coherent and are nontrivial. The main weakness is the β>1/2 branch of Theorem 4.1, which is used to claim a sharp threshold and is supported by an unjustified contour estimate.","major_comments":[{"comment":"The estimate |⟨χ,e^{-itH_L(μ_L)}P_Lχ⟩−e^{-itE0}|≤C_4 L^{1−2β} is asserted to be uniform in t from the contour representation, but this does not follow. On γ(L) with Im z>0, |e^{-itz}|=e^{t Im z}=e^{O(1)t}; at t=τ μ_L^{-2}=τ L^{2β} this factor is enormous. Moreover, P_Lχ is not an eigenvector with eigenvalue E0, so replacing e^{-itH_L}P_Lχ by e^{-itE0} is unjustified. In fact, for μ_L≪d(ν_L,ν∞), second-order perturbation theory gives an eigenvalue shift μ_L^2 Re Σ_L(E0)+O(μ_L^4), so at t=τ μ_L^{-2} the amplitude acquires a phase e^{-iτ Re Σ_L(E0)} → e^{iΔ(E0)τ}, not 1. Thus the claimed β>1/2 branch and the 'sharp threshold' statement in the introduction are unsupported and appear incorrect.","section":"Theorem 4.1, β∈(1/2,∞) case, after Eq. (4.6)"},{"comment":"The stated rate d^{min{2βα_-,(1−2β)α_-}} does not follow from Theorem 1.1 for all β<1/2 and all α_-<α. Substituting μ_L=d^β and t=τ μ_L^{-2} into (1.11) with γ=α_- gives the factor exp(τ d^{1−2β(1+α_-)}). This diverges unless 1−2β(1+α_-)≥0, i.e. β≤1/(2(1+α_-)). For β close to 1/2 and α_- not sufficiently small, the estimate is not a consequence of the stated theorem. The existence part of the limit can be recovered by taking γ small, but the 'moreover' rate claim and its use in Theorem 4.1 require a separate argument or a restricted parameter range.","section":"Corollary 1.1 and Theorem 4.1, β∈(0,1/2) rate"}],"minor_comments":[{"comment":"Notation for the Fourier transform is inconsistent: the text writes \\hat p_ε(t) but the displayed equation uses bpε(t); the decay e^{-ε|t|} should be stated as the Fourier transform of the Cauchy kernel.","section":"Eq. (1.16) and surrounding text"},{"comment":"Spell 'Poincaré' consistently; the current 'Poincare' appears in a proof relying on the standard inequality.","section":"Lemma 4.1"},{"comment":"The sequence ε_L is introduced with values in (1/2,3/2) but the notation is then reused for the contour γ(L); clarify that the radius is ε_L and avoid conflict with the smearing scale ε from Section 3.","section":"Theorem 4.1"}],"recommendation":"major_revision","confidential_remarks":"The central Theorem 1.1 and the existence claim of Corollary 1.1 appear defensible, and Lemma 4.1 is a solid, useful quantitative result. However, the β>1/2 branch of Theorem 4.1 is not merely a gap in presentation; the claimed limit 1 is inconsistent with second-order perturbation theory, which predicts a phase e^{iΔτ}. The 'sharp threshold' claim is therefore likely wrong at the level of the value of the limit, not just the proof. The authors should either correct Theorem 4.1 to the expected e^{iΔτ} (with a valid proof) or remove the sharp-threshold claim. The rate claim in Corollary 1.1 also needs qualification. These are load-bearing but appear fixable within the scope of the paper, hence major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main result of this paper is the approximate exponential decay theorem for sequences of Friedrichs Hamiltonians with discrete spectra, with error controlled by Lévy distance. That part looks right to me. The proof via Cauchy-smearing and self-energy regularity is coherent, and the paper is careful to note the error has to grow in time because of Poincaré recurrences. The extension of Jensen–Nenciu to the finite-volume setting is genuinely new, and the spin-boson application with the β<1/2 condition is a clean corollary. The derivation of the spectral density for the cavity model is also a solid piece of work.\n\nThe soft spot is the β>1/2 case of Theorem 4.1. The argument chooses a contour γ(L) of radius O(1), bounds |Σ_L(z)|≤CL, then claims a t-uniform bound of order L^{1−2β} for the contour integral. That does not follow. On the arc with Im z>0, |e^{-itz}|=e^{t Im z} grows in t, so pointwise control of the integrand gives nothing uniform in t. And the contour encloses real axis points of σ(K_L), so you cannot deform to the lower half-plane without picking up residues with t-dependent phases. My reading is that the no-decay threshold β=1/2 is unsupported. The stress-test note correctly identifies this. It doesn't affect Theorem 1.1 or Corollary 1.1, but the 'sharp threshold' claim should be fixed or removed.\n\nTwo smaller things. Theorem 1.1 states 'for all μ≥0', but μ=0 is not covered because d(ν_L,ν_∞)/μ^2 is undefined; restricting to μ>0 is harmless. Also, the proof of Theorem 1.1 only imposes d/μ^2<q0, while the useful regime for exponential decay on times μ^{-2} needs d/μ^{2+2γ} small; the stated error bound is fine because the exponential factor captures it, but it is worth stating the timescale explicitly.\n\nOverall, this is a serious paper. The main theorem and the cavity density computation deserve a referee. The β>1/2 claim should not be accepted without a fix. I would send it to review with a request to repair or remove that part.","headline":"The finite-volume exponential decay theorem is solid and worth refereeing; the claimed sharp β=1/2 threshold is not proven — the t-uniform bound in the β>1/2 case does not follow from the contour estimate.","tokens_in":19422,"tokens_out":2307,"would_cite":true,"duration_ms":25970,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81Q10","81Q15","47A55"],"pacs":[],"model":"deepseek-v4-flash","headline":"For Friedrichs Hamiltonians with purely discrete bath spectra, exponential decay at the Fermi golden rule rate emerges in the weak-coupling limit provided the level spacing is far smaller than the resonance width µ².","keywords":["Friedrichs Hamiltonians","exponential decay","Fermi's golden rule","discrete spectrum","Lévy distance","weak-coupling limit","survival probability","spin-boson model"],"falsifier":"Take a Friedrichs Hamiltonian satisfying (Aα), choose µ_L = d(ν_L,ν_∞)^{0.4}, and compute the survival amplitude at rescaled time τ=1 (i.e., t=1/µ_L²) for increasing L; the theorem says this converges to exp(iΔ(E0)-Γ(E0)) with Γ(E0)=πρ_∞(E0). If the computed limit deviates from this, the central claim is wrong.","tokens_in":18520,"feed_emoji":"⚛️","tokens_out":14297,"duration_ms":136489,"temperature":0.7,"pith_summary":"Physically, an excited atom should decay even when its radiation field is confined to a finite cavity, where the spectrum is a discrete ladder that becomes denser as the cavity grows. This paper proves that for the class of rank-one Friedrichs Hamiltonians such decay is indeed exponential on the Fermi time scale: the survival probability is close to e^{-2µ²Γt} with Γ the Fermi golden rule rate computed from the limiting spectral density, provided the discrete levels are much more closely spaced than the coupling-induced resonance width. The proof yields a quantitative error bound controlled by the Lévy distance between the discrete spectral measure and its continuous limit, and it identifies the scale separation d ≪ ε ≪ µ² required for the approximation to work. A concrete application shows that a two-level atom in a large cavity, in the rotating-wave approximation, undergoes exponential decay when the coupling decreases slower than L^{-1/2} as the cavity grows.","feed_headline":"Exponential decay emerges from dense discrete spectra","feed_subtitle":"When a cavity's level spacing is far below the coupling width, excited states decay at the Fermi rate.","key_machinery":"The argument turns on the self-energy Σ_L(E+iε)=⟨g_L,(K_L-E-iε)^{-1}g_L⟩ and the exact identity for the Cauchy-smeared spectral density: (p_ε * σ_μ^L)(E) = (1/π) Im[(E0 - E - iε - µ²Σ_L(E+iε))^{-1}]. This reduces the problem to controlling two regularity facts: the self-energy is Hölder continuous up to an error d/ε, and its value approaches Δ(E0)+iπρ_∞(E0) with error d/ε + ε^α. The smearing scale ε is chosen to be d µ^{-2γ}, which lies in the window d ≪ ε ≪ µ². In that window the smeared density is close in L¹ to a Lorentzian with centre E0-µ²Δ and width ε+µ²Γ, whose Fourier transform is the exponential decay law. The distribution of scales — spacing d, smearing ε, resonance width µ² — is t","core_discovery":"The central claim is a quantitative theorem: for sequences of Friedrichs Hamiltonians whose bath spectral measures converge weakly to an absolutely continuous measure with Hölder continuous density near E0, the survival amplitude equals e^{-i(E0-µ²Δ(E0))t} e^{-µ²Γ(E0)t} plus an error bounded by C(γ)(µ^{2γ} + (d/µ²)^α) exp(d/µ^{2γ} t), where Γ(E0)=πρ_∞(E0) is the Fermi golden rule rate, Δ(E0) the energy shift, and d the Lévy distance between discrete and limiting measures. If d ≪ µ², the error is small up to times ~µ^{-2}, and in the simultaneous weak-coupling/continuum limit the decay becomes exactly exponential. The proof uses a Cauchy-smearing scale ε with d ≪ ε ≪ µ² to turn the discrete s","pith_inferences":["This suggests a practical design rule for cavity experiments: exponential decay is observable only when the coupling-induced resonance width exceeds the typical cavity level spacing, so tuning µ or L to enter this regime should be possible.","The unprobed borderline β=1/2, where level spacing and resonance width are of the same order, likely hosts a crossover to non-exponential decay; numerical study of this boundary could reveal universal scaling.","The Cauchy-smearing method could be pushed beyond leading order: keeping higher-order terms in the self-energy would give quantitative corrections to the decay law and energy shift at stronger coupling, and the same machinery might handle multi-level atoms.","Since only the Lévy distance controls the error, the result should extend to disordered or quasiperiodic level sequences whose distribution functions converge uniformly; testing such sequences would probe how robust the emergent exponential decay is."],"forward_implications":["For the confined spin-boson model with coupling µ_L = L^{-β}, exponential decay with rate Γ(E0)=πρ_∞(E0) occurs for β<1/2; for β>1/2 the survival amplitude instead tends to 1 (the state fails to decay).","The error bound grows exponentially in time, implying that for any finite cavity the exponential law must eventually fail — a quantitative signature of Poincaré recurrences.","The theorem recovers, as a special case, the known uniform-in-time estimates for Friedrichs Hamiltonians with absolutely continuous spectrum, and extends them to discrete-but-dense spectra.","The result supplies a finite-volume justification of the Fermi golden rule for the confined spin-boson model and, by the same mechanism, for other finite-volume models of metastability before any thermodynamic limit is taken."],"fun_headline_variants":["Discrete spectra mimic continuum: Fermi decay emerges","Weak coupling to dense levels yields exponential decay","Cavity QED: exponential decay from discrete modes","Discrete bath, exponential decay: Fermi rule survives","From discrete to continuous: emergent exponential decay"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The argument stands or falls on the assumption that the discrete spectral measures approach a Hölder-regular absolutely continuous measure fast enough that the typical level spacing d is much smaller than the coupling-induced resonance width µ²; if this scale separation fails, the error term (d/µ²)^α is not small and the Lorentzian approximation breaks down.","fun_headline_variants_meta":{"raw":{"variants":["Discrete spectra mimic continuum: Fermi decay emerges","Weak coupling to dense levels yields exponential decay","Cavity QED: exponential decay from discrete modes","Discrete bath, exponential decay: Fermi rule survives","From discrete to continuous: emergent exponential decay"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000161,"raw_usage":{"total_tokens":1067,"prompt_tokens":735,"completion_tokens":332,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":479,"completion_tokens_details":{"reasoning_tokens":260}},"tokens_in":479,"tokens_out":332,"duration_ms":3445,"temperature":1.0,"reasoning_tokens":260,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T06:11:25.692058+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a Friedrichs Hamiltonian satisfying (Aα), choose µ_L = d(ν_L,ν_∞)^{0.4}, and compute the survival amplitude at rescaled time τ=1 (i.e., t=1/µ_L²) for increasing L; the theorem says this converges to exp(iΔ(E0)-Γ(E0)) with Γ(E0)=πρ_∞(E0). If the computed limit deviates from this, the central claim is wrong.","supporting_citations":[],"review_version":1}