{"id":"5abff1e1-0d05-4265-a9f8-6ad7c57235b8","arxiv_id":"2411.13972","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"In the high-temperature limit, the rescaled eigenvalues of the stochastic Bessel operator converge to a new, non-Poissonian point process built from alternating reflected Brownian motions.","lead":"This paper calculates what happens to the tiny energy levels of a random operator, the stochastic Bessel operator, as the temperature goes to infinity: after a clever rescaling, they settle into a new random pattern. The result is unexpected because high temperature usually scrambles eigenvalues into independent noise, but here a repulsive structure survives.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"SDE (5) does not follow from (3): Itô's formula gives noise β dW and a divergent 1/(2β) drift, so the limiting reflected Brownian motion is not the limit of qβ_μ as stated.","rationale":"The reader's weakest-assumption identification (Proposition 2) is valid: the existence and uniqueness of M^0 is asserted rather than proved. However, the more load-bearing point is the derivation of the rescaled SDEs (5)-(6), the object of the entire proof. If (5) does not follow from (3), then the limiting reflected Brownian motion r+ is not the limit of q+, and every estimate in Section 4 and the tightness argument in Section 5 is addressing the wrong dynamics. I checked the Itô computation twice; the discrepancy is a leading-order factor 1/β in the drift and β in the noise, so it cannot be dismissed as o(1). This is an internal inconsistency between equation (3), quoted from [4], and equation (5), stated as its consequence. It may be a typo (the coefficient 2√β in (3) should probably be 2/√β, which would make (5) exact), but as written the proof of Theorem 1 is unverified. Since the central claim may still be true after correction, I do not recommend outright rejection, but the paper cannot be assessed in its current form; hence UNVERDICTED. The reader's CONDITIONAL verdict should be tightened until the SDE derivation is fixed and a proof of Proposition 2 is supplied.","tokens_in":13219,"tokens_out":27892,"duration_ms":248973,"concrete_test":"Re-derive (5) from (3) symbolically: set X_t = p_{Λβ}(t/(4β)), W_t = 2√β B(t/(4β)), q = β ln X, and compute dq via Itô's formula. Record the coefficients of dt and dW. If the dt coefficient contains +1/(2β) (or -β/2) and the dW coefficient is β, then (5) is refuted. Alternatively, correct (3) to coefficient 2/√β and recompute Definition 1's operator to see which version is the actual stochastic Bessel operator.","verdict_should_be":"UNVERDICTED","load_bearing_attack":"Section 3.2 defines q+_μ(t)=β ln p_{Λβ}(t/(4β)), with p following (3). Let W(t)=2√β B(t/(4β)); this W is a standard Brownian motion. The rescaled process X_t=p_{Λβ}(t/(4β)) satisfies dX = X dW + ((a+2/β)X - X^2 - e^{-μ/β}e^{-t/(4β)}) dt/(4β). Itô's formula then gives d ln X = dW + ((a+2/β)-X-e^{-μ/β}e^{-t/(4β)}/X) dt/(4β) - dt/2. Multiplying by β yields dq+ = β dW + [a/4 + 1/(2β) - X/4 - (1/4)e^{-(q+ + t/4 + μ)/β} - β/2] dt. Compared with (5), the noise is β dW rather than dW, and the drift contains an additional +1/(2β)-β/2 term, which is leading-order as β→0. Therefore (5) is not a consequence of (3); it is compatible only if the Brownian coefficient in (3) is 2/√β rather than 2√β. If (3) is as written, the diffusions q± have vanishing noise and a divergent drift, so the convergence to the reflected Brownian motions in Definition 2 (Proposition 5) is not established, and the limiting process in Theorem 1 is unsupported.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the low-lying eigenvalues of the stochastic Bessel operator G^{β,a} in the high-temperature limit β→0. After the rescaling μ^β(k)=β ln(1/Λ^{β,a}(k)), it claims that the point process of rescaled eigenvalues converges to a random simple point process governed by a diffusion r_μ that alternates between reflected Brownian motions with drifts a/4 and −(a+1)/4, restarting at 0 whenever it hits the critical line c_μ(t)=−μ−t/4. The proof is based on a Riccati transform and coupled diffusions q^±_μ, with convergence statements in Propositions 3–5 and tightness in Section 5. The main theorem is Theorem 1, which depends on Propositions 2 and 4.","tokens_in":13512,"tokens_out":14287,"duration_ms":130823,"significance":"If correct, the result would give a concrete high-temperature limit for the hard edge of the β-Laguerre spectrum, showing that the limiting point process retains repulsion and differs from the Poisson limit obtained for the stochastic Airy operator. The proposed characterization by alternating reflected Brownian motions is natural and interesting, and the scaling β ln(1/λ) is well motivated. However, the paper's central derivation of the rescaled SDEs is incorrect, and the limiting object is therefore not connected to the stochastic Bessel operator as written. The paper also leaves Proposition 2 essentially unproved and relies on sketched estimates in Proposition 5. These are not presentation issues but load-bearing gaps, so the central claim is currently unsupported.","major_comments":[{"comment":"The Itô computation leading to (5) is not correct. Let X_t = p_{Λβ}(t/(4β)) and W_t = 2√β B_{t/(4β)}, which is a standard Brownian motion. From (3), dX_t = X_t dW_t + [(a+2/β)X_t − X_t^2 − e^{−μ/β}e^{−t/(4β)}]/(4β) dt. Since q^+_μ = β ln X_t, Itô's formula gives d q^+_μ = β dW_t + [a/4 + 1/(2β) − β/2 − (1/4)e^{q^+_μ/β} − (1/4)e^{−(q^+_μ+μ+t/4)/β}]dt. This is not (5): the Brownian coefficient is β, not 1, and the drift contains the divergent term 1/(2β). An analogous computation for q^−_μ gives a Brownian coefficient β and an additional positive 1/(2β) drift, so (6) is also not a consequence of (3). Since Proposition 5 and Definition 2 are both based on (5)–(6), the convergence to the reflected Brownian motions described in Theorem 1 is not established from the stochastic Bessel equation.","section":"Section 3.2, Eqs. (5) and (6)"},{"comment":"Proposition 2 is asserted rather than proved. The sentence preceding it, 'Since µ ∈ R+ ↦ ν^0_µ(R+) decreases from ∞ to 0 almost surely, it is easy to prove...', does not demonstrate the three properties that make M^0 well-defined: finiteness of ν^0_µ(R+) for each fixed µ, monotonicity in µ, and uniqueness of the resulting discrete set. These properties are exactly what allows the finite-dimensional relations M^0[µ_i,∞[ = ν^0_{µ_i}(R+) to define a point process. Because Proposition 4 identifies the limits of the eigenvalue point process with M^0, this gap is load-bearing for Theorem 1.","section":"Section 3.3, Proposition 2"},{"comment":"The proof of Proposition 5, which is the only bridge from the SDEs to the limiting process, is sketched at several crucial points. In §4.2.2, Lemma 1 is stated for any γ<0, but the text chooses γ=−c(T). Since c(T)=−μ−T/4, this γ is positive, so Lemma 1 does not apply. If γ=c(T) was intended, one still needs an argument that hitting the constant level c(T) before l_2 implies hitting the moving barrier c(t) before l_2. The passages around (17)–(18) ('it suffices to show') and the bracketing inequalities (19)–(20) are also asserted rather than derived. These are not minor omissions: without a complete proof of Proposition 5, the identification of the limiting measure ν^0_µ in Proposition 3 is unsupported.","section":"Section 4, Proposition 5 and §4.2.2"}],"minor_comments":[{"comment":"The text 'wether it prevails' should read 'whether it prevails'.","section":"Section 3.2"},{"comment":"The region described as 'between −c_µ(t)+δ0 and −δ0' appears to be c_µ(t)+δ0; as written the lower endpoint is μ+t/4+δ0, which is not compatible with the bound (24).","section":"Section 4, Lemma 2"},{"comment":"In the paragraph after (25), the notation T is used for a fixed horizon while T_ϵ and T_0 are also present; explicitly distinguishing these would help the reader.","section":"Section 5"}],"recommendation":"reject","confidential_remarks":"The manuscript is within scope for math.PR and the topic is timely. The difficulty is technical rather than scholarly: the Itô calculation in Section 3.2 invalidates the proposed limiting process as derived from the SBO. I see no indication of duplicate publication or citation manipulation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe short version: this paper has a genuinely interesting idea — the high-temperature limit of the stochastic Bessel operator eigenvalues, after rescaling mu = beta ln(1/lambda), should be a non-Poissonian point process built from alternating reflected Brownian motions. But the central derivation of the SDEs that define the limiting object is off by a factor of beta in the noise and misses a divergent 1/(2beta) drift. As written, the main theorem is unsupported.\n\nWhat's actually new: the rescaling is natural and the claim that the limit is not Poisson (unlike the stochastic Airy case of Dumaz-Labbé) is worth taking seriously. The proof strategy — Riccati transform, explosion times of the rescaled diffusions, tightness of the explosion measure — is coherent and the paper is clearly organized. The literature citation looks fair.\n\nThe problem: Section 3.2 starts from (3), dp = 2√β p dB + ((a+2/β)p − p² − λ e^{−t})dt, and defines q+ = β ln p(t/(4β)). Setting W(t)=2√β B(t/(4β)), the Itô formula gives\n\ndq+ = β dW + [a/4 + 1/(2β) − e^{q+/β}/4 − e^{−(q+ + t/4 + μ)/β}/4 − β/2] dt,\n\nnot (5). The noise is β dW, not dW, and the 1/(2β) drift is leading order as β→0. The same issue affects (6). So the diffusions q± do not converge to the reflected Brownian motions of Definition 2; Proposition 5 and Theorem 1 don't follow.\n\nThis is load-bearing, not a typo in a final formula: the entire limiting process is defined through (5)-(6). If the coefficient 2√β in (3) were 2/√β, the calculation would work, but that's not what the paper says, and the SBO definition in [12] uses 2√β.\n\nThe reader's report mentioned sketchiness in Proposition 2, Lemma 1, and Proposition 5 — those are real but secondary. The SDE error is the first thing a referee should check.\n\nBottom line: the idea is worth pursuing, but this version has a fatal gap in its core derivation. I'd send it to review so the error is documented and the authors have a chance to fix the rescaling or correct the model, but I wouldn't accept it in anything close to current form.","headline":"Interesting non-Poissonian limit claim, but the core SDE derivation in Section 3.2 is off by a factor of beta in the noise and a missing 1/(2beta) drift, so Theorem 1 is unsupported as written.","tokens_in":14067,"tokens_out":7176,"would_cite":false,"duration_ms":65600,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H25","60B20","60J60"],"pacs":[],"model":"deepseek-v4-flash","headline":"The stochastic Bessel operator's low eigenvalues converge, after $\\beta\\ln(1/\\lambda)$ rescaling, to a random point process driven by alternating reflected Brownian motions.","keywords":["stochastic Bessel operator","Beta-Laguerre ensemble","high-temperature limit","point process convergence","Riccati transform","reflected Brownian motion","hard edge"],"falsifier":"Compute, for a fixed $a>0$ and several $\\mu$, the number of restarts of the limiting diffusion $r_\\mu$ before it stops hitting the line $c_\\mu$, and compare that count with the number of rescaled eigenvalues $\\mu_\\beta(k)\\ge\\mu$ obtained by numerically solving the stochastic Bessel operator at small $\\beta$. If the counts do not match in law for small $\\beta$, or if the number of restarts is infinite with positive probability for some $\\mu$, Theorem 1 fails.","tokens_in":12982,"feed_emoji":"🎲","tokens_out":8667,"duration_ms":68687,"temperature":0.7,"pith_summary":"The paper proves that the low-lying eigenvalues of the stochastic Bessel operator, after the rescaling $\\mu_\\beta(k)=\\beta\\ln(1/\\Lambda_{\\beta,a}(k))$, converge in law as $\\beta\\to0$ to a random simple point process on $\\mathbb R_+$. The limit is characterized by a single alternating diffusion: reflected Brownian motion with drift $a/4$, restarted at $0$ whenever it hits the line $-\\mu-t/4$, then reflected Brownian motion with drift $-(a+1)/4$, and so on. Unlike the high-temperature limit of the stochastic Airy operator, the limiting process is not Poisson; repulsion between points survives the $\\beta\\to0$ limit. The result gives a concrete description of the high-temperature hard-edge spectrum and opens the way to studying its statistics.","feed_headline":"Bessel spectrum survives high temperature as a repulsive point process","feed_subtitle":"The rescaled eigenvalues converge to a coupled-diffusion limit, not a Poisson process.","key_machinery":"The Riccati transform $p=\\psi'/\\psi$ converts the eigenvalue equation $G^{\\beta,a}\\psi=\\Lambda\\psi$ into the coupled diffusions $p^\\beta_\\lambda$ of equation (3). The identity (4), which equates the event that $p^\\beta_\\lambda$ explodes at most $k$ times with the event $\\Lambda_{\\beta,a}(k)>\\lambda$, is the bridge from explosions to eigenvalues. After rescaling $\\lambda=\\exp(-\\mu/\\beta)$ and time $t/(4\\beta)$, the logarithmic transformation yields the alternating diffusions $q^\\pm_\\mu$ with SDEs (5) and (6). As $\\beta\\to0$ their limit is $r_\\mu$, built from reflected Brownian motions with drifts $a/4$ and $-(a+1)/4$; the critical line $c_\\mu(t)$ governs when the alternation occurs. This mechanism carries the proof: controlling the first explosion times (Proposition 5) and proving tightness of the explosion-time measures (Section 5) yields Proposition 3 and then Theorem 1.","core_discovery":"The central claim is Theorem 1: for fixed $a>0$, as $\\beta$ tends to $0$, the point process $(\\mu_\\beta(k), k\\ge0)$ with $\\mu_\\beta(k)=\\beta\\ln(1/\\Lambda_{\\beta,a}(k))$ converges in law, in the left-vague/right-weak topology on measures on $\\mathbb R_+$, to a random simple point process. The limit is defined through the coupled measures $\\nu^0_\\mu$: for each $\\mu>0$, $\\nu^0_\\mu$ counts the times $\\xi^-_0(i)$ at which the limiting diffusion $r_\\mu$ hits the critical line $c_\\mu(t)=-\\mu-t/4$, where $r_\\mu$ alternates between reflected Brownian motions with drifts $a/4$ and $-(a+1)/4$ built from a single Brownian motion. The paper proves convergence of the explosion-time measures $\\nu^\\beta_\\mu$ to $\\nu^0_\\mu$, derives the finite-dimensional marginals $M^\\beta[\\mu,\\infty)$ to $M^0[\\mu,\\infty)=\\nu^0_\\mu(\\mathbb R_+)$, and then uses tightness to conclude the process convergence. The construction of $M^0$ via Proposition 2 is asserted rather than proved, and the convergence theorem is directed at exactly this limiting object.","pith_inferences":["A natural next step is to extend the rescaling to $a\\in(-1,0]$, which the paper explicitly excludes; the drift signs and the critical-line mechanism would likely need modification.","If the convergence can be pushed through the known hard-edge connection with the beta-Laguerre ensemble, the same limiting point process would describe the high-temperature, high-dimensional hard edge of Wishart-type spectra.","The coupling structure suggests a direct simulation route: run a single Brownian path, build $r_\\mu$ for a grid of $\\mu$, and count hits of the critical line, giving a concrete estimator of $M^0$ to compare with SBO simulations."],"forward_implications":["The rescaled low-lying SBO spectrum has a non-Poisson limit: the limiting point process is simple and retains a repulsive structure even as $\\beta\\to0$.","For each $\\mu>0$, the number of limiting eigenvalues above $\\mu$ equals $\\nu^0_\\mu(\\mathbb R_+)$, the number of hits of the critical line by the alternating reflected Brownian motion $r_\\mu$.","The finite-dimensional marginals converge jointly: for fixed $\\mu_1<\\dots<\\mu_k$, the vector of counts $(M^\\beta[\\mu_1,\\infty),\\dots,M^\\beta[\\mu_k,\\infty))$ converges in law to $(\\nu^0_{\\mu_1}(\\mathbb R_+),\\dots,\\nu^0_{\\mu_k}(\\mathbb R_+))$.","The same Brownian motion drives all the measures $\\nu^0_\\mu$, so the limiting point process is defined through a single coupling rather than independent noises."],"supporting_citations":[{"why":"Introduces the stochastic Bessel operator and proves the explosion count identity (4) linking the number of explosions of $p^\\beta_\\lambda$ to the eigenvalues of $G^{\\beta,a}$.","marker":"[12]"},{"why":"Supplies the SDE system for $(\\psi,\\psi')$ and the Riccati diffusion (3) that the rescaling starts from.","marker":"[4]"},{"why":"Provides the high-temperature limit of the stochastic Airy operator, whose Poisson limit is the contrast that frames the repulsive limit found here.","marker":"[3]"},{"why":"Supplies the Riccati transform $p=\\psi'/\\psi$ used to turn the eigenvalue equation into coupled diffusions.","marker":"[9]"},{"why":"Gives the Kallenberg compactness criterion used to prove tightness of the explosion-time measures.","marker":"[10]"}],"fun_headline_variants":["High-temp Bessel eigenvalues: coupled diffusions, not Poisson","Bessel high-temp limit: repulsive point process from coupled SDEs","Stochastic Bessel at high temp: eigenvalues converge to coupled limit","High temperature Bessel operator: coupled-diffusion point process","Bessel spectrum hot limit: repulsive point process via coupled diffusions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the proposed limiting object $M^0$ exists: Proposition 2 states that $\\nu^0_\\mu(\\mathbb R_+)$ is finite for every $\\mu>0$, decreases from infinity to zero as $\\mu$ grows, and yields a unique discrete process, but these properties are asserted as easy to prove rather than demonstrated.","fun_headline_variants_meta":{"raw":{"variants":["High-temp Bessel eigenvalues: coupled diffusions, not Poisson","Bessel high-temp limit: repulsive point process from coupled SDEs","Stochastic Bessel at high temp: eigenvalues converge to coupled limit","High temperature Bessel operator: coupled-diffusion point process","Bessel spectrum hot limit: repulsive point process via coupled diffusions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000177,"raw_usage":{"total_tokens":1266,"prompt_tokens":888,"completion_tokens":378,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":504,"completion_tokens_details":{"reasoning_tokens":284}},"tokens_in":504,"tokens_out":378,"duration_ms":4129,"temperature":1.0,"reasoning_tokens":284,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:40:46.798520+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, for a fixed $a>0$ and several $\\mu$, the number of restarts of the limiting diffusion $r_\\mu$ before it stops hitting the line $c_\\mu$, and compare that count with the number of rescaled eigenvalues $\\mu_\\beta(k)\\ge\\mu$ obtained by numerically solving the stochastic Bessel operator at small $\\beta$. If the counts do not match in law for small $\\beta$, or if the number of restarts is infinite with positive probability for some $\\mu$, Theorem 1 fails.","supporting_citations":[{"cited_title":"and Rider, B","cited_arxiv_id":null,"evidence_quote":"Introduces the stochastic Bessel operator and proves the explosion count identity (4) linking the number of explosions of $p^\\beta_\\lambda$ to the eigenvalues of $G^{\\beta,a}$."},{"cited_title":"and V alkó, B","cited_arxiv_id":null,"evidence_quote":"Supplies the SDE system for $(\\psi,\\psi')$ and the Riccati diffusion (3) that the rescaling starts from."},{"cited_title":"and Labbé, C","cited_arxiv_id":null,"evidence_quote":"Provides the high-temperature limit of the stochastic Airy operator, whose Poisson limit is the contrast that frames the repulsive limit found here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Riccati transform $p=\\psi'/\\psi$ used to turn the eigenvalue equation into coupled diffusions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Kallenberg compactness criterion used to prove tightness of the explosion-time measures."}],"review_version":1}