{"id":"fe105d84-e30c-4056-b24a-7c000b1b67f9","arxiv_id":"2608.06646","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For the 3D stochastic wave equation with noise white in time and colored in space, a unique global mild solution exists when drift and diffusion grow at most like |u| (log |u|)^θ, with θ below explicit thresholds.","lead":"This paper proves that the three-dimensional stochastic wave equation with multiplicative colored Gaussian noise has a unique global solution, even when the drift and diffusion terms grow slightly faster than linearly. It extends known one- and two-dimensional results to the harder three-dimensional case, where the wave kernel is a measure supported on a sphere.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 3.1's proof uses an unjustified |σ(0)|/L_σ ≤ 1 bound; the main theorem relies on this proposition, but the gap is patchable.","rationale":"The central stopping-time argument in Section 4 appears to close, provided Propositions 3.1 and 3.2 hold. The technical core is the Hölder regularity of the truncated solutions. I found one concrete flaw: in the proof of Proposition 3.1, the estimate of Q21 assumes |σ(0)|/L_σ is bounded by a constant, which is not a hypothesis of the proposition. This is a genuine gap in the proof as written, but it is easily patched because the theorem only uses the proposition for the truncated coefficients σ_N with L_σN growing, so the ratio is bounded for large K. The reader's verdict of CONDITIONAL is therefore appropriate. The reader's stated weakest assumption, the sphere-integral conditions (1.9)-(1.10) in Assumption 1.1, is a valid concern about the theorem's scope but not an internal inconsistency: the theorem is conditional on these assumptions, and they are verified for three natural classes of kernels. If forced to identify the single most load-bearing internal issue, it is the Proposition 3.1 proof gap rather than the restrictiveness of Assumption 1.1. Hence partial agreement with the reader: they flagged the same gap in their rationale but selected a different weakest assumption.","tokens_in":28471,"tokens_out":35413,"duration_ms":283116,"concrete_test":"Recompute the Q21 step in the proof of Proposition 3.1 (around Eq. (3.8)) replacing the bound |σ(0)|/L_σ ≤ 1 by the sharp bound |σ(u)|^2 ≤ (|σ(0)| + L_σ |u|)^2. Verify whether the extra term |σ(0)|^2 |w|^{2γ1} ∫_0^t (t-s)^{ν1} ds can be bounded by C M e^{cβt} |w|^{2γ̄} with M from (2.6). If yes, the proposition is valid as stated; if not, confirm that adding L_σ ≥ 2|σ(0)| to Proposition 3.1 and checking L_σN ≥ 2|σ(0)| for the truncated coefficients in Section 4 restores the proof of Theorem 1.2.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Proposition 3.1, the estimate for Q21 in (3.8) is obtained by writing |σ(u)|^2 ≤ L_σ^2 (1 + |u|^2), which implicitly assumes |σ(0)|/L_σ ≤ 1. The proposition is stated for arbitrary globally Lipschitz σ, and no such lower bound on L_σ is assumed. If |σ(0)| ≫ L_σ, the σ(0) contribution to Q21 is of order |σ(0)|^2 |w|^{2γ1} ∫_0^t (t-s)^{ν1} ds, which is not controlled by the term involving sup ||u||_p^2 and must be absorbed into the right-hand side of (3.1). The proof as written cannot be followed without this extra bound. In Section 4, the proposition is applied to truncated coefficients σ_N with L_σN ~ (log N)^{θ2} → ∞ as N grows, so the ratio |σ(0)|/L_σN is bounded for N large; nevertheless, the statement of Proposition 3.1 is not uniform for all globally Lipschitz σ unless the constant is allowed to depend on |σ(0)|/L_σ. The fix is to either add the hypothesis L_σ ≥ C|σ(0)| to the proposition, or carry the |σ(0)|^2 terms through and absorb them into M in (2.6). Since Theorem 1.2 only uses the proposition with large L_σN, this gap does not invalidate the main theorem, but it renders the proof of Proposition 3.1 incomplete.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the three-dimensional stochastic wave equation (1.1) driven by multiplicative Gaussian noise that is white in time and correlated in space, with locally Lipschitz drift b and diffusion coefficient σ satisfying the logarithmic growth conditions (1.11). Under Assumption 1.1 on the spatial covariance f, the authors prove existence and uniqueness of a global mild solution on any fixed time interval [0,T] (Theorem 1.2). The proof combines a truncation/stopping-time argument adapted from Mueller [11] with spatial and temporal Hölder regularity estimates for the equation with globally Lipschitz coefficients (Propositions 3.1 and 3.2). These regularity estimates rely on Assumption 1.1, which is verified for C^2_b functions, Riesz kernels, and Bessel kernels in Section 5. The paper also gives a self-contained moment bound (Proposition 2.1) for the globally Lipschitz case.","tokens_in":28802,"tokens_out":19314,"duration_ms":162416,"significance":"If the results are correct, this is a meaningful extension of Mueller's long-time existence theory for stochastic wave equations from dimensions one and two to dimension three, where the wave kernel is a surface measure and the analysis is considerably more delicate. The paper makes the covariance assumptions explicit and demonstrates them on several natural examples, and the proof is presented in considerable detail. The main technical gap identified in Proposition 3.1 is local and appears readily fixable without changing the theorem; the other issues are notational or typographical. The reliance on [9] is as a tool for regularity estimates, and I do not see a circularity problem.","major_comments":[{"comment":"The estimate of Q21 in (3.8) is obtained by writing |σ(u)|^2 ≤ L_σ^2(1+|u|^2), which uses the implicit inequality |σ(0)| ≤ C L_σ. This is not a hypothesis of Proposition 3.1, which is stated for arbitrary globally Lipschitz σ. The sentence 'In Q21 we have bounded |σ(0)|/Lσ by a constant since finally in this paper Lσ will be large, see (4.2)' refers to the truncated coefficients in Section 4 and does not justify the general statement. The proof can be repaired either by adding L_σ ≥ C|σ(0)| as an explicit hypothesis of Proposition 3.1, or by carrying the extra |σ(0)|^2 |w|^{2γ1} ∫_0^t (t-s)^{ν1} ds term through the Gronwall argument and absorbing it into the forcing term M|w|^{2γ̄} on the right-hand side of (3.1). Since Theorem 1.2 uses the proposition only for truncations with L_{σ_N} = O((log N)^{θ_2}) → ∞, this gap does not invalidate the main theorem, but the proof as written is incomplete.","section":"Section 3.1, Eq. (3.8)"},{"comment":"The exponent in the statement is displayed as '0< µ < µ= 1/2 min{2µ2, ν+ 1,2 γ, γ+µ 1}' with γ undefined, while the proof uses '0<2µ <2 µ := min{2µ2,2 γ, ν+ 1,γ+µ 1}'. The symbol γ should be ¯γ = min{α1,α2,α3,γ1,γ2}, and the two displays should be normalized so that the Hölder exponent claimed in (3.13) is unambiguous. This matters because Lemma 4.4 and Proposition 4.2 subsequently use the exponent γ∧µ with a separate parameter γ ∈ (0,γ̄), and the current notation makes it difficult to verify that the constants are uniform over the truncation levels and over the choice of γ.","section":"Section 3.2, Proposition 3.2 and proof"},{"comment":"The event E_1 is defined as {sup_{C(t_1,0)} |u^{N_1}(t,x)| ≤ M + K^2}, but the induction in (4.9) and the argument in Lemma 4.3 require the base case sup_{C(t_1,0)} |u^{N_1}(t,x)| ≤ M + K/2, which is the value obtained by substituting n=1 into the general increment K 2^{n-2}. With the printed definition, E_1 only gives M+K^2, and the chain of inequalities in (4.9) fails for large K; the proof of Proposition 4.2 is therefore not justified as written. This is evidently a typographical slip (K^2 should be K/2), but it must be corrected, or the induction in (4.9) reformulated, before the stopping-time proof is valid.","section":"Section 4, definition of E_1 (Eq. (4.7))"}],"minor_comments":[{"comment":"The verifications of (1.9) and (1.10) for the Riesz and Bessel kernels are delegated to [9, Proposition 5.3] and [9, Proposition 5.4] without explicitly listing the inherited parameter choices for γ1, γ2, µ1, µ2, ν1, and ν2. Please state the resulting parameter assignments so that the reader can confirm that the version of Assumption 1.1 used here is indeed satisfied.","section":"Section 5.2 and 5.3"},{"comment":"The growth condition (1.11) is written with log|z|, while the abstract uses log_+(z)=log(z∨e). For consistency, define log_+ immediately before (1.11) or assume |z| large enough so that the two statements agree outside a compact set.","section":"Theorem 1.2 and abstract"},{"comment":"The constant M in (2.6) contains T∥v0∥_{L∞} + ∥u0∥_{L∞} + T∥∇u0∥_{L∞}; since the preceding display bounds |V(t,x)| by t∥v0∥_{L∞} + ∥u0∥_{L∞} + t∥∇u0∥_{L∞}, it would be clearer to write sup_{t∈[0,T]} explicitly when defining M.","section":"Proposition 2.1, Eq. (2.6)"}],"recommendation":"major_revision","confidential_remarks":"The first author of this manuscript is also a co-author of [9], from which several key estimates are taken. I examined the use of [9] and find it non-circular: the cited results concern regularity and covariance estimates for equations with Lipschitz coefficients, and the present theorem does not reduce to them. The gap in Proposition 3.1 and the typo in (4.7) are both local and easily repairable. The paper is within the scope of a probability journal and, after a careful revision, would likely be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nShort version: this is a genuine new theorem—existence and uniqueness for the 3D stochastic wave equation with multiplicative noise and coefficients with logarithmic superlinear growth—and the proof is mostly solid. It deserves a serious referee, but Proposition 3.1 has a real hole that needs patching before the paper is complete.\n\nWhat's new: Mueller's 1997 result covered R and R^2 with σ(u) up to (log|u|)^α, α<1/2. Huang and Tao push this to R^3, where the wave kernel is a surface measure, which is the technically awkward case. They introduce Assumption 1.1 on the spatial covariance (second-order spherical cancellation estimates), prove spatial and temporal Hölder moment bounds, then close the argument with the standard truncation-and-stopping-time scheme. The parameter threshold θ2 < (ν̄+1)/2 is explicit and natural. The verification for C^2_b, Riesz, and Bessel kernels is concrete and useful. This is real work, and the main theorem is not a corollary of anything in [9]—the earlier paper is used for lemmas, which is legitimate.\n\nSoft spots. The stress-test concern is real: in the proof of Proposition 3.1, the estimate for Q21 uses |σ(u)|² ≤ L_σ²(1+|u|²) without assuming |σ(0)|/L_σ ≤ C. As stated, the proposition is for arbitrary globally Lipschitz σ, so the proof as written doesn't go through. The authors acknowledge this in a parenthetical (\"since finally in this paper L_σ will be large, see (4.2)\") but that doesn't fix the statement. The fix is easy—either add the hypothesis or carry the |σ(0)|² term through and absorb it into M—and because the truncation levels in Section 4 have L_σN growing like (log N)^θ2, the main theorem is probably safe. But the proposition needs to be stated correctly.\n\nSecond, the paper leans heavily on [5], [7], [9] for key estimates. That's not a flaw by itself, but it does mean the referee should check those imports carefully. Third, the assumptions on initial data are a bit stiff (u0 ∈ C^2, bounded, with Hölder Δu0), but that matches the 3D kernel and isn't a problem.\n\nBottom line: this is a solid technical contribution for the SPDE regularity community. Send it to a serious referee, but expect a revision. If the authors fix Prop 3.1 and make the dependence on [9] explicit, I'd be happy with it.","headline":"Genuine extension of Mueller's wave-equation result to 3D with a new covariance assumption; the main theorem is new, but Proposition 3.1 has an unpatched hypothesis gap that should be fixed before publication.","tokens_in":29318,"tokens_out":3962,"would_cite":true,"duration_ms":31665,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H15","35L05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the three-dimensional stochastic wave equation with multiplicative Gaussian noise admits a unique global mild solution when drift and diffusion grow only logarithmically faster than linearly, under covariance…","keywords":["stochastic wave equation","global mild solution","non-Lipschitz coefficients","multiplicative Gaussian noise","spatial covariance","Hölder regularity","Riesz kernel","Bessel kernel"],"falsifier":"Compute the left-hand sides of (1.9) and (1.10) for a spatial covariance $f$ that satisfies (1.6)--(1.8) but has oscillatory or anisotropic structure, such as $f(z)=\\cos(k\\cdot z)/(1+|z|^2)$; if either quantity fails to be $O(h^{\\mu_1})$ or $O(h^{2\\mu_2})$ with positive exponents as $h\\to 0$, then Proposition 3.2's temporal H\\\"older estimate is unavailable, and the paper's proof of Theorem 1.2 does not close for that noise.","tokens_in":28260,"feed_emoji":"🌊","tokens_out":12466,"duration_ms":113812,"temperature":0.7,"pith_summary":"This paper establishes global existence and uniqueness for the three-dimensional stochastic wave equation driven by multiplicative Gaussian noise that is white in time and colored in space, when the drift and diffusion coefficients are only locally Lipschitz and grow logarithmically faster than linearly. The result covers $b(u)=u\\log_+u^{\\theta_1}$ with $\\theta_1\\in(0,2)$ and $\\sigma(u)=u\\log_+u^{\\theta_2}$ with $\\theta_2\\in(0,(\\bar\\nu+1)/2)$, where $\\bar\\nu$ is fixed by the spatial covariance of the noise. A sympathetic reader should care because most previous work on non-Lipschitz stochastic PDEs concerned heat equations, while the wave equation in three dimensions is harder: its fundamental solution is a measure supported on a sphere. The proof handles that sphere by recentering the surface measures and exploiting second-order cancellation estimates for the covariance, then uses a stopping-time argument adapted from the one- and two-dimensional theory.","feed_headline":"Superlinear drift and noise still yield global wave solutions in 3D","feed_subtitle":"The threshold on the log-growth exponents is explicit, and covers Riesz and Bessel noise covariances.","key_machinery":"The central object is the 3D wave kernel $G_t=\\frac{1}{4\\pi t}A_t$, where $A_t$ is uniform surface measure on the sphere of radius $t$, with the scaling property $G_t(dy)=tG_1(t^{-1}dy)$ and the convolution identity $(G_s*G_t)(dx)=\\frac{1}{8\\pi|x|}\\mathbf{1}_{[s-t,s+t]}(|x|)\\,dx$. The argument recenters all spherical measures at the origin to turn stochastic-integral terms into convolutions, where Assumption 1.1's bounds (1.6)--(1.8) control the $L^p$ moments and the spherical cancellation estimates (1.9)--(1.10) control temporal increments. These H\\\"older estimates feed a stopping-time construction: drift and diffusion are truncated at levels $N_n=K2^n$, and events on dyadic space-time grids ensure the truncated solutions agree up to a common time that grows to infinity.","core_discovery":"The central claim is Theorem 1.2: for a spatial covariance $f$ satisfying Assumption 1.1 (in particular the integrability and spherical-difference bounds (1.6)--(1.10)), and for locally Lipschitz drift $b$ and diffusion $\\sigma$ with $b(z)=O(|z|(\\log|z|)^{\\theta_1})$, $\\sigma(z)=O(|z|(\\log|z|)^{\\theta_2})$, $0<\\theta_1<2$, $0<\\theta_2<(\\bar\\nu+1)/2$ with $\\bar\\nu=\\min\\{\\nu,\\nu_1,\\nu_2\\}$, there exists a unique global mild solution to (1.1) on every $[0,T]$ from bounded H\\\"older initial data. The mechanism is to prove moment bounds and spatial and temporal H\\\"older estimates for the truncated globally Lipschitz equations, then to patch solutions across an increasing sequence of stopping times using the logarithmic growth to keep the patching events overwhelmingly likely until $\\tau_\\infty=\\infty$.","pith_inferences":["Going beyond the paper: the explicit threshold $\\theta_2<(\\bar\\nu+1)/2$ raises the question of whether $\\theta_2=(\\bar\\nu+1)/2$ is a genuine critical exponent; at that value the summability estimate in Proposition 4.2 stops being strict, and testing the borderline case would require a separate argument.","Going beyond the paper: the spherical cancellation conditions (1.9)--(1.10) are shape conditions on the covariance rather than simple integrability conditions, so covariances with strong anisotropy or oscillation may fail them even when (1.6)--(1.8) hold; checking such an $f$ would delineate the method's scope.","Going beyond the paper: the same machinery of H\\\"older estimates plus stopping times could be adapted to other wave-type equations with measure-supported kernels, such as the damped wave equation in three dimensions, where the kernel has similar sphere structure."],"forward_implications":["For any fixed $T>0$ and admissible initial data, the solution $u$ exists, is unique, and has finite $L^p$ moments uniformly on $[0,T]\\times\\mathbb{R}^3$.","The theorem applies to Riesz kernels $f(z)=|z|^{-\\beta}$ with $0<\\beta<2$ and to Bessel kernels of order $\\alpha>3$, giving explicit ranges for $\\theta_1$ and $\\theta_2$ in both cases.","It also applies to every bounded covariance $f\\in C^2_b(\\mathbb{R}^3)$, with the simplest parameter values $\\nu=\\nu_1=\\nu_2=2$ and $\\gamma_1=\\gamma_2=\\mu_1=\\mu_2=1$.","Within the stated growth classes, no finite-time blow-up can occur on any fixed time interval, in contrast to supercritical polynomial growth where earlier blow-up results apply.","The mild solution is built coherently from truncated equations, so it coincides with the unique Lipschitz-coefficient solution on every set where the solution stays below the truncation level."],"supporting_citations":[{"why":"Supplies the H\\\"older-regularity framework for the 3D wave equation with sphere-supported kernel, and the verification of Assumption 1.1 for Riesz and Bessel kernels.","marker":"[9]"},{"why":"Provides the stopping-time and truncation argument that the paper adapts to prove global existence for non-Lipschitz coefficients.","marker":"[11]"},{"why":"Supplies the recentering and change-of-variables technique for spherical surface measures and a deterministic regularity lemma used in the H\\\"older estimates.","marker":"[5]"},{"why":"Sets up the It\\^o--Walsh stochastic integral framework and the existence and uniqueness theory for globally Lipschitz coefficients.","marker":"[4]"},{"why":"Establishes the globally Lipschitz mild-solution theory for the nonlinear stochastic wave equation in any dimension, including the integrability condition (1.3).","marker":"[15]"},{"why":"Supplies the Burkholder--Davis--Gundy inequality used throughout the moment and increment estimates.","marker":"[10]"}],"fun_headline_variants":["Log-superlinear drift: 3D wave equation has unique global solution","3D stochastic wave: log growth still yields global solutions","Beyond Lipschitz: 3D wave PDE admits unique global mild solution","Log growth doesn't stop waves: global uniqueness in 3D SWE","Superlinear coefficients: global solution for 3D stochastic wave"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof stands on the two spherical cancellation estimates (1.9) and (1.10) for the spatial covariance $f$: without them the temporal H\\\"older bound in Proposition 3.2 does not follow, and the stopping-time argument cannot close; these estimates are verified for smooth, Riesz, and Bessel kernels but are not consequences of the simpler growth bounds (1.6)--(1.8).","fun_headline_variants_meta":{"raw":{"variants":["Log-superlinear drift: 3D wave equation has unique global solution","3D stochastic wave: log growth still yields global solutions","Beyond Lipschitz: 3D wave PDE admits unique global mild solution","Log growth doesn't stop waves: global uniqueness in 3D SWE","Superlinear coefficients: global solution for 3D stochastic wave"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000986,"raw_usage":{"total_tokens":4212,"prompt_tokens":1003,"completion_tokens":3209,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":619,"completion_tokens_details":{"reasoning_tokens":3115}},"tokens_in":619,"tokens_out":3209,"duration_ms":20896,"temperature":1.0,"reasoning_tokens":3115,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T04:11:17.325224+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the left-hand sides of (1.9) and (1.10) for a spatial covariance $f$ that satisfies (1.6)--(1.8) but has oscillatory or anisotropic structure, such as $f(z)=\\cos(k\\cdot z)/(1+|z|^2)$; if either quantity fails to be $O(h^{\\mu_1})$ or $O(h^{2\\mu_2})$ with positive exponents as $h\\to 0$, then Proposition 3.2's temporal H\\\"older estimate is unavailable, and the paper's proof of Theorem 1.2 does not close for that noise.","supporting_citations":[{"cited_title":"Partial Differ","cited_arxiv_id":null,"evidence_quote":"Supplies the H\\\"older-regularity framework for the 3D wave equation with sphere-supported kernel, and the verification of Assumption 1.1 for Riesz and Bessel kernels."},{"cited_title":"Dalang and Marta Sanz-Sol´ e,H¨ older-Sobolev regularity of the solution to the stochastic wave equation in dimension three, Mem","cited_arxiv_id":null,"evidence_quote":"Supplies the recentering and change-of-variables technique for spherical surface measures and a deterministic regularity lemma used in the H\\\"older estimates."},{"cited_title":"Dalang and Llu ´ ıs Quer-Sardanyons,Stochastic integrals for spde’s: a comparison, Expo","cited_arxiv_id":null,"evidence_quote":"Sets up the It\\^o--Walsh stochastic integral framework and the existence and uniqueness theory for globally Lipschitz coefficients."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the globally Lipschitz mild-solution theory for the nonlinear stochastic wave equation in any dimension, including the integrability condition (1.3)."},{"cited_title":"119, Published for the Conference Board of the Mathematical Sciences, Washington, DC; by the American Mathematical Society, Providence, RI, 2014","cited_arxiv_id":null,"evidence_quote":"Supplies the Burkholder--Davis--Gundy inequality used throughout the moment and increment estimates."}],"review_version":1}