{"id":"445ca570-c2bf-4ac9-8da8-7f4df7862b59","arxiv_id":"2501.19362","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Infrared-divergent spin boson model loses its ground state above a finite critical coupling, proved via long range order in a dual continuum Ising model.","lead":"The authors prove that the infrared-divergent spin boson model has no ground state when the coupling is large, and therefore undergoes a phase transition in the coupling strength. The proof translates the quantum problem into a one-dimensional continuum Ising model and uses percolation theory to show that this model has long range order at strong coupling.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the reader's φ-map counterexample uses a continuous interpolation that the paper's half-open piecewise-linear extension does not define; under the stated map the inequality and the percolation comparison go through.","rationale":"The reader's weakest-assumption analysis identifies the φ-map inequality as the key risk, but the stated counterexample computes φ using a continuous linear interpolation between consecutive integer values, which is not the map defined in the paper. The paper defines φ piecewise with slope +1 on [2n,2n+1) and slope −1 on [2n+1,2n+2); these half-open intervals allow a discontinuous bijection onto R\\{−1,0}. Under that map the inequality is true, and the subsequent stochastic domination of the two-sided percolation by the one-sided model has a uniform constant because the image of the integer lattice under φ expands distances by at least a factor 1/2 in the worst case. The |n−m|=2 exclusion is precisely the pair where the image intervals become adjacent, and the proof supplies a separate condition p_{0,2}(α)>p_{0,1}(β) for that case. I checked the main chain of implications: Theorem 3.4 gives the vacuum-overlap expansion, Corollary 3.5 turns long-range order into absence of ground states, Lemma 4.3 connects spin correlations to continuum percolation, and Theorem 4.1 establishes long-range order under the stated slow decay of g. No internal inconsistency or unjustified step of comparable weight emerged. The paper would benefit from a few clarifying sentences about the discontinuous nature of φ and the site-reflection in the edge-probability comparison, but these are presentation issues rather than load-bearing defects.","tokens_in":22664,"tokens_out":54760,"duration_ms":521969,"concrete_test":"Recompute p_{n,m}(α) and p_{φ(n),φ(m)}(β) for g(t)=C(1+t²)^{-1} with the half-open φ defined above, verifying numerically for representative pairs (0,1), (0,3), (2,5), (100,101), and (100,102) that there exists a uniform c<∞, for example c=16, with p_{φ(n),φ(m)}(α/c)≤p_{n,m}(α) for all n≠m, and that |φ(t)−φ(s)|≥|t−s|/4 for s=2.5, t=5.5 and for random samples from intervals with |⌊s⌋−⌊t⌋|≠2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The weakest spot in the paper is the one-sided/two-sided percolation comparison in Theorem 4.1, but the reader's specific objection does not land. The map is defined by φ(t)=t−n on [2n,2n+1) and φ(t)=n−t on [2n+1,2n+2), with half-open intervals; it is discontinuous at integers and is a bijection from [0,∞) onto R\\{−1,0}. For s=2.5 and t=5.5, φ(2.5)=1.5 and φ(5.5)=−3.5, so |φ(t)−φ(s)|=5.0≥0.75. A direct case check confirms |φ(t)−φ(s)|≥|t−s|/4 whenever |⌊s⌋−⌊t⌋|≠2. The subsequent edge comparison reduces to F(|φ(n)−φ(m)|)/F(|n−m|)≤c with c≈4 for g=(1+t²)^{-1}, since the folding map has integer-Lipschitz constant 1/2; the |n−m|=2 case is handled separately by p_{0,2}(α)>p_{0,1}(β). I do not find a load-bearing error in the central argument: Theorem 3.4, Corollary 3.5, Lemma 4.3, and the percolation comparison together support Theorem 2.9. Minor presentational gaps, such as the exact role of site-reflection in the edge comparison, are repairable and do not threaten the claim.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves absence of ground states for the infrared-divergent spin boson model at large coupling, and combines this with known small-coupling existence results to establish a unique phase transition in the coupling strength. The main technical steps are: an exact expansion of the vacuum overlap ρ(λ) in terms of correlation functions of a one-sided continuum Ising model (Theorem 3.4); a criterion showing that long-range order in that Ising model implies ρ(λ)=0 when v/ω is not square-integrable (Corollary 3.5); and a proof of long-range order at large coupling using a continuum percolation representation and a comparison between one-sided and two-sided site-bond percolation models (Theorem 4.1). The central result is Theorem 2.9, which states that under the infrared decay condition ∫ e^{-tω}|v|² dk ≥ C(1+t²)^{-1}, the critical coupling λ₀ is finite.","tokens_in":23005,"tokens_out":50212,"duration_ms":466191,"significance":"If correct, the paper resolves a conjecture in the spectral theory of the spin boson model and gives the first rigorous proof of a coupling-driven ground-state phase transition in this model. The vacuum overlap expansion, the derivation of the identity ρ=lim Z²/Z, and the detailed proof of convergence from discrete to continuum percolation are valuable technical contributions in their own right. The paper is self-contained modulo standard external results in percolation theory and prior work on existence criteria for ground states, and it also provides an explicit upper bound on the vacuum overlap. The central strategy is coherent and the main claims are supported by the argument as written.","major_comments":[],"minor_comments":[{"comment":"The inequality |φ(t)-φ(s)| ≥ |t-s|/4 is essential for the comparison of the one-sided and two-sided percolation models, but it is asserted without proof. Because the extension φ is discontinuous at integers, this is not immediate; please add a short case check or a reference. In particular, for the half-open definition φ(t)=t-n on [2n,2n+1) and φ(t)=n-t on [2n+1,2n+2), the example s=2.5, t=5.5 gives φ(s)=1.5 and φ(t)=-3.5, so the inequality holds; nevertheless, a proof should be supplied to rule out other cases.","section":"Section 4, proof of Theorem 4.1"},{"comment":"The constant c in p_{φ(n),φ(m)}(α/c) ≤ p_{n,m}(α) is not specified. For g(t)=C/(1+t²), the pointwise bound |φ(t)-φ(s)| ≥ |t-s|/4 yields g(φ(t)-φ(s)) ≤ 16 g(t-s), so c=16 suffices; this should be written out explicitly.","section":"Section 4, around Eq. (4.11)"},{"comment":"The vertex-alive comparison is not stated among the large-α conditions. Besides α/c>β and p_{0,2}(α)>p_{0,1}(β), one also needs p₀(α)≥p₀(β); since p₀(α)↑1 as α→∞, this is harmless, but it should be included.","section":"Section 4, around Eqs. (4.7)-(4.11)"},{"comment":"The notation P_{α,T,N} is used both for the FK measure and for the independent percolation measure in (4.2)-(4.5); this overloaded notation makes the argument harder to follow. Distinct symbols such as P^{FK} and P^{ind} would be clearer.","section":"Lemma 4.3 and Proposition A.5"},{"comment":"The subscript 'T N' in P_{α,T,T N}(0↔n) appears to be a typo for P_{α,T,N}(0↔n).","section":"Appendix A, Proposition A.5"},{"comment":"The displayed relative bound is missing arguments on the right-hand side: the terms ε||dΓ(A)|| and ε^{-1}||(1+A^{-1/2})|| should act on ψ.","section":"Section 2.1, Eq. (2.1)"},{"comment":"The equality ~E_{α,2T}[∏ X_{s_i}X_{t_i}] = ~E_{α,2T}[∏ X_{s_i}X_{t_i}|X_T=1] uses spin-flip symmetry; a sentence explaining this step would improve readability.","section":"Section 3, after Eq. (3.5)"},{"comment":"The application of [NS86] to the site-bond model is terse. Please spell out that one can first choose the long-range edge probabilities large enough, then take p₀ and p_{0,1} close to 1, so that the existence of a percolating β follows by monotonicity.","section":"Section 4, footnote 1"}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is technically sound in my reading. The most delicate part is the percolation comparison in Theorem 4.1; although the argument is correct, it is presented too tersely in places (the φ-map inequality, the constant c, the vertex-alive condition, and the use of [NS86]). I recommend that the authors expand these points during revision to make the proof easier to verify. No concerns about novelty or scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is the real thing. The paper proves the finite-critical-coupling conjecture for the infrared-divergent spin boson model: above a finite λ0 there is no ground state, and combined with the known small-coupling existence this gives a unique phase transition. The two genuinely new pieces are the vacuum overlap expansion (Theorem 3.4) that expresses the ground-state overlap as an inverse of a sum over convolutions of Ising n-point functions, and the percolation proof of long range order in the one-sided continuum Ising model at large coupling (Theorem 4.1). Both are coherent and, as far as I can see, correct.\n\nThe proof of Theorem 4.1 is the part that will get scrubbed in referee reports. The folding map φ that compares the one-sided site-bond percolation on N0 with the two-sided model on Z is introduced in one sentence, and the inequality |φ(t)-φ(s)| ≥ |t-s|/4 appears without proof. The reader's counterexample (s=2.5, t=5.5) uses the naive continuous interpolation; with the paper's half-open linear definition, φ(2.5)=0.5 and φ(5.5)=-3.5, so the difference is 4.0 ≥ 0.75. A direct case check shows the inequality holds whenever |⌊s⌋-⌊t⌋|≠2, and the |n-m|=2 case is handled separately. So the objection does not land. The real soft spot is presentation: the map is discontinuous at every integer, and the site-aliveness comparison quietly assumes the percolation on [0,∞) is translation invariant for n≥1. None of this threatens the argument.\n\nThe paper is careful about the one-sided vs two-sided distinction, which genuinely matters for long-range models, and the appendix proves the discrete-to-continuum percolation convergence that Spohn left out. The citations are on point; self-citations are for technical lemmas that are either proved in the text or genuinely external.\n\nWho should read it: anyone working on non-perturbative QFT, infrared problems, or the spin-boson model. It's a serious result, not a small incremental step. I'd send it to a qualified referee. Recommend peer review.","headline":"The finite-coupling conjecture for the infrared-divergent spin boson model is proved; the reader's φ-map counterexample does not survive contact with the actual definition, and the paper deserves serious refereeing.","tokens_in":23542,"tokens_out":10117,"would_cite":true,"duration_ms":86881,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81T10","82B20","82B26","60K35"],"pacs":[],"model":"deepseek-v4-flash","headline":"The infrared-divergent spin boson model has no ground state once the coupling exceeds a finite critical value, establishing a phase transition in the coupling strength.","keywords":["spin boson model","ground state existence","infrared divergence","phase transition","continuum Ising model","long range order","percolation","vacuum overlap"],"falsifier":"Check the comparison inequality used in the proof of Theorem 4.1: for the map $\\varphi$ defined on $\\mathbb{N}_0$ and extended linearly to $[0,\\infty)$, test whether $|\\varphi(t)-\\varphi(s)|\\ge|t-s|/4$ for all $s,t\\ge0$ with $|\\lfloor s\\rfloor-\\lfloor t\\rfloor|\\ne2$. The specific pair $s=2.5$, $t=5.5$ gives $|\\varphi(t)-\\varphi(s)|=0.5<0.75$, so the bound as stated fails and the stochastic domination step needs a modified comparison; alternatively, a direct simulation of $\\tau_{\\alpha,1}(t)$ for $g(t)=(1+t^2)^{-1}$ at large $\\alpha$ would show whether long range order actually holds.","tokens_in":22463,"feed_emoji":"🧲","tokens_out":13763,"duration_ms":113610,"temperature":0.7,"pith_summary":"This paper proves that the spin boson model—a two-state quantum system coupled to a massless scalar field—has no ground state when the spin–field coupling is large enough, provided the interaction is infrared-divergent in the precise sense that the kernel $g(t)=\\int e^{-|t|\\omega(k)}|v(k)|^2dk$ decays no faster than $(1+t^2)^{-1}$. Combined with known results that a ground state exists for small coupling, this establishes a phase transition in the coupling strength. The proof converts the question into one about long range order in a dual one-dimensional continuum Ising model: the vacuum overlap, which is positive exactly when a ground state exists, is shown to be the inverse of a sum of Ising correlation functions, and long range order makes that sum diverge. The paper also shows the phase transition is unique, meaning there is a single critical coupling constant below which a ground state exists and above which none does.","feed_headline":"Strong coupling destroys the ground state in spin boson model","feed_subtitle":"Ground states exist only below a finite critical coupling, shown via long-range order in a dual Ising model.","key_machinery":"The central object is the continuum Ising model on the half-line: the law of a $\\{-1,1\\}$-valued continuous-time random walk on $[0,T]$ weighted by $\\exp(\\alpha\\int\\int_{[0,T]^2}g(t-s)X_sX_t\\,dsdt)$, with $g$ given by the Fourier decay kernel and $\\alpha=\\lambda^2/8$. The load-bearing identity is the vacuum-overlap expansion of Theorem 3.4, which expresses $\\rho(\\lambda)$ as the inverse of a convolution sum of Ising $n$-point functions; this identity turns ground-state existence into convergence of that sum. The long-range-order proof then proceeds by discretizing the continuum model into a lattice Ising model, passing to the FK-percolation representation, and using stochastic domination to compare the one-sided percolation model on $\\mathbb{N}_0$ with a two-sided model on $\\mathbb{Z}$ that is known to have an infinite cluster at large coupling.","core_discovery":"The central claim is Theorem 2.9: if $\\int_{\\mathbb{R}^d} e^{-t\\omega(k)}|v(k)|^2dk \\ge C(1+t^2)^{-1}$ for some $C>0$ and all $t>0$, then the critical coupling $\\lambda_0$ is finite, so $H_\\lambda$ has no ground state for all $|\\lambda|>\\lambda_0$. The mechanism is the vacuum-overlap identity $\\rho(\\lambda)=\\left(\\sum_{n\\ge0}\\frac{(2\\alpha)^n}{n!}\\int_{[0,\\infty)^n}(\\tau_{\\alpha,n}*\\tau_{\\alpha,n})(t)\\prod_{i=1}^n g(t_i)\\,dt_i\\right)^{-1}$ with $\\alpha=\\lambda^2/8$, where $\\tau_{\\alpha,n}$ are $n$-point functions of a half-line continuum Ising model. Long range order in that model, $\\inf_{t\\ge0}\\tau_{\\alpha,1}(t)>0$, forces the $n=1$ term to diverge because $v/\\omega\\notin L^2$ under the assumed decay. Since $\\rho(\\lambda)>0$ is equivalent to the existence of a ground state, this proves absence at large coupling. The same expansion yields monotonicity of $\\rho$ in $|\\lambda|$ and hence a unique critical $\\lambda_0$.","pith_inferences":["Beyond the paper, the same percolation-based route may apply to translation-invariant polaron models, where absence of ground states at large total momentum is conjectured; the correlation-function criterion developed here is a plausible template for such proofs.","If the phase transition in the one-sided continuum Ising model is sharp in the sense of finite susceptibility below the critical $\\alpha$, then the critical coupling $\\lambda_0$ for the spin boson model would coincide with the KMS phase-transition point found in earlier work; the paper notes the missing connection, and the main obstacle is likely the boundary effects in the long-range model.","Since the paper omits explicit numerical values for $\\lambda_0$, a testable extension is to compute the two-point function $\\tau_{\\alpha,1}(t)$ in Monte Carlo simulations for $g(t)=(1+t^2)^{-1}$; the onset of long range order would give an estimate of the critical $\\alpha$, hence of $\\lambda_0$."],"forward_implications":["There exists a finite critical coupling $\\lambda_0$ such that $H_\\lambda$ has a ground state for $|\\lambda|<\\lambda_0$ and none for $|\\lambda|>\\lambda_0$; the value at $\\lambda_0$ itself is left open.","The vacuum overlap obeys the explicit upper bound $\\rho(\\lambda)\\le\\exp(-\\frac{\\lambda^2}{4}\\int_0^\\infty\\int_{\\mathbb{R}^d}t|v(k)|^2e^{-t(\\omega(k)+2)}dk\\,dt)$, so the would-be ground state becomes macroscopically bosonic at large coupling.","For the standard example $\\omega(k)=|k|$, $|v(k)|\\sim|k|^{-\\delta}$, the theorem applies exactly in the infrared-divergent regime $\\delta\\in[\\frac{d}{2}-1,\\frac{d}{2}-\\frac12)$, including the physically relevant case $d=3$, $\\delta=\\frac12$.","Sufficient criteria for existence and absence of ground states are unified in terms of Ising correlation functions: finite integrated susceptibility gives existence, long range order gives absence."],"supporting_citations":[{"why":"Supplies the percolation strategy for long range order in a continuum Ising model and the comparison argument that this paper adapts to empty boundary conditions on a half-line.","marker":"[Spo89]"},{"why":"Proves existence of a percolation transition for one-dimensional long-range models with $|i-j|^{-s}$, $s\\le2$, giving the infinite cluster on the two-sided lattice used in Theorem 4.1.","marker":"[NS86]"},{"why":"Provides the FK-percolation representation (Lemma 2.1) and stochastic domination theorem (Theorem 4.1) that connect Ising correlations to percolation probabilities.","marker":"[ACCN88]"},{"why":"Introduced the discretization scheme for the continuum Ising model that the paper uses to take scaling limits.","marker":"[SD85]"},{"why":"Proves the scaling limit of the discretized continuum Ising model and the correlation bound behind Proposition 2.8.","marker":"[HHS22a]"},{"why":"Establishes the small-coupling existence of ground states in the infrared-divergent spin boson model, the other side of the phase transition.","marker":"[HH11]"}],"fun_headline_variants":["Spin boson ground state dies at finite critical coupling","No ground state above critical coupling in spin boson model","Critical coupling separates ground states in spin boson model","Ground state in spin boson model only below finite coupling"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The comparison between the one-sided and two-sided percolation models needs the distance bound $|\\varphi(t)-\\varphi(s)|\\ge|t-s|/4$ for all admissible pairs; if that inequality fails, the stochastic domination argument that produces long range order collapses.","fun_headline_variants_meta":{"raw":{"variants":["Spin boson ground state dies at finite critical coupling","No ground state above critical coupling in spin boson model","Critical coupling separates ground states in spin boson model","Ground state in spin boson model only below finite coupling"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000533,"raw_usage":{"total_tokens":2583,"prompt_tokens":983,"completion_tokens":1600,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":599,"completion_tokens_details":{"reasoning_tokens":1535}},"tokens_in":599,"tokens_out":1600,"duration_ms":10559,"temperature":1.0,"reasoning_tokens":1535,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T20:25:05.642887+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check the comparison inequality used in the proof of Theorem 4.1: for the map $\\varphi$ defined on $\\mathbb{N}_0$ and extended linearly to $[0,\\infty)$, test whether $|\\varphi(t)-\\varphi(s)|\\ge|t-s|/4$ for all $s,t\\ge0$ with $|\\lfloor s\\rfloor-\\lfloor t\\rfloor|\\ne2$. The specific pair $s=2.5$, $t=5.5$ gives $|\\varphi(t)-\\varphi(s)|=0.5<0.75$, so the bound as stated fails and the stochastic domination step needs a modified comparison; alternatively, a direct simulation of $\\tau_{\\alpha,1}(t)$ for $g(t)=(1+t^2)^{-1}$ at large $\\alpha$ would show whether long range order actually holds.","supporting_citations":[],"review_version":1}