{"id":"178bb379-509a-45b7-aafc-8b4040207386","arxiv_id":"1908.03944","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper establishes local and global well-posedness for the 2D stochastic heat and damped wave equations with exponential nonlinearity in the ranges β²<1.37π (heat, any sign), β²<4π (heat, defocusing), and β²<0.86π (wave, defocusing), with invariance of the associated Gibbs measures.","lead":"Mathematicians proved when two-dimensional stochastic heat and wave equations with an exponential force have unique solutions, extending earlier results to a wider range of the coupling constant. The main novelty is a new way to split the wave equation's rough part from its smoother part, which also resolves an open problem in the field.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.9's proof is conditional on an omitted Bourgain invariant-measure step, and Section 6.2's truncated wave system has a residual-initial-data mismatch that must be repaired.","rationale":"The reader’s conditional verdict is appropriate. The local and global heat results (Theorems 1.1, 1.2, 1.6) are supported by detailed estimates: moment bounds on the Gaussian multiplicative chaos via Brascamp-Lieb and Kahane, the sign-definite structure, compactness arguments, and a careful finite-dimensional generator computation for the truncated invariance. The local wave theorem (1.8/1.15) is also carried through with explicit parameter choices in (6.8) and contraction estimates. The weak point is confined to the final hyperbolic Gibbs theorem: Bourgain’s invariant measure argument is invoked without details, and Section 6.2 as written does not set up the truncated dynamics consistently. Specifically, (6.25) uses initial data (Q_Nw0,Q_Nw1), while the decomposition uses Ψ_wave with initial data (w0,w1), so the residual does not start at zero and the system (6.26) omits its linear evolution. This is likely fixable—one can take Ψ_N with Q_N initial data and pass to the limit, or replace the initial data in (6.25) by (w0,w1), exploiting the density bound R_N≤1 to transfer convergence from μ1⊗μ0 to ρwave,N. But the fix is not in the paper. Since the missing Bourgain step is standard and the local estimates are solid, this does not warrant rejection; it warrants conditional acceptance pending a written verification of the global-in-time argument and correction of the truncated setup.","tokens_in":66619,"tokens_out":22169,"duration_ms":214026,"concrete_test":"Re-derive the residual system for (6.25) at t=0: compute (u_N−Ψ_wave)(0) with u_N(0)=Q_Nw0 and Ψ_wave(0)=w0. If this is nonzero, then either replace the initial data in (6.25) by (w0,w1) so that Ψ_wave matches the initial data, or add the missing linear evolution term to (6.26); then check that the estimates of Section 6.2 (IV1, IV2, and Step 3) still yield convergence of the truncated solutions to the full flow. This single analytical check determines whether the issue is a harmless typo or a genuine missing argument.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 6.2 proves Theorem 1.9 only by reference: “The rest of the argument follows from applying Bourgain’s invariant measure argument [9,10]. Since it is standard, we omit details.” This is the load-bearing step for almost sure global well-posedness and invariance of ρwave, and it is not demonstrated for this specific truncated damped wave dynamics. Moreover, there is a concrete setup inconsistency: (6.25) is stated with initial data (Q_Nw0,Q_Nw1), but the decomposition u_N=X_N+Y_N+Ψ uses Ψ=Ψ_wave from (1.36), which has initial data (w0,w1). The residual (X_N+Y_N)(0) then equals (Q_N−Id)w0, and its time derivative is nonzero as well, whereas the system (6.26) defines X_N,Y_N by Duhamel integrals with zero initial data. Thus the truncated system solved in (6.26) is not literally the one whose invariant measure is ρwave,N in (1.22). The gap is likely repairable—one can choose the stochastic convolution with Q_N initial data, or add the small linear evolution of (Q_N−Id)w0—but as written the passage from (6.25) to (6.26) is not proven. Combined with the omitted global-existence and Borel-Cantelli details of Bourgain’s argument, Theorem 1.9 is conditional on a standard but unverified adaptation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops a local and global solution theory for the two-dimensional stochastic nonlinear heat equation (1.1) and the stochastic damped nonlinear wave equation (1.2), both with an exponential nonlinearity λβe^{βu} and additive space-time white noise. In the parabolic case it proves local well-posedness for general λ≠0 and 0<β²<8π/(3+2√2), global well-posedness for λ>0 and β²<4π, and invariance of the renormalized Gibbs measure for λ>0 in that range. In the hyperbolic case it proves local well-posedness for λ>0 and 0<β²<(32−16√3)π/5≈0.86π, and asserts almost sure global well-posedness and Gibbs invariance for the same β-range. The arguments combine the Da Prato–Debussche trick with new moment bounds for Gaussian multiplicative chaos obtained by Brascamp–Lieb and Kahane inequalities, a sign-definite structure in the defocusing case, and an X+Y decomposition of the residual wave dynamics.","tokens_in":66845,"tokens_out":5590,"duration_ms":49911,"significance":"If fully established, these results constitute a substantial advance: they extend Garban's range with noise continuity, give the first Gibbs-invariant dynamics for the damped wave equation with exponential nonlinearity in this β-regime, and answer the Sun–Tzvetkov open question for α=1. The paper is unusually detailed in the deterministic estimates: Proposition 1.12 is proved by two independent methods, the threshold computations in §§4 and 6.1 are explicit, and the local statements in Theorems 1.1, 1.2, and 1.8 are supported by self-contained proofs. However, the hyperbolic global result (Theorem 1.9) is not proved in the text: it depends on an omitted Bourgain invariant-measure argument and, more concretely, on an initial-data mismatch between the truncated system under study and the system whose invariant measure is ρ_wave,N. The parabolic global result (Theorem 1.6) is fully argued and does not share this gap.","major_comments":[{"comment":"The truncated system analyzed in (6.26) is not literally the system (6.25) whose invariant measure is ρ_wave,N. Indeed, (6.25) is stated with initial data (Q_N w0, Q_N w1), while the decomposition u_N = X_N + Y_N + Ψ uses the untruncated stochastic convolution Ψ from (1.36), whose initial data are (w0, w1). Consequently (X_N + Y_N)(0) = (Q_N − Id)w0 and ∂_t(X_N + Y_N)(0) = (Q_N − Id)w1, whereas the Duhamel integrals in (6.26) define X_N and Y_N with zero initial data. This is a concrete inconsistency, not merely a cosmetic one: the well-posedness and convergence statements proved for (6.26) do not automatically transfer to the dynamics (6.25) started from the truncated Gibbs measure. The gap appears repairable, for example by using a truncated stochastic convolution Ψ_N with initial data (Q_N w0, Q_N w1), or by adding the linear evolution of (Q_N − Id)(w0, w1) to the residual, but as written the proof of Theorem 1.9 does not cover the stated dynamics.","section":"§6.2, Eqs. (6.25)–(6.26) and (1.36)"},{"comment":"Theorem 1.9's almost sure global well-posedness and invariance of ρ_wave rest on the sentence 'The rest of the argument follows from applying Bourgain's invariant measure argument [9,10]. Since it is standard, we omit details.' This step is load-bearing: it must produce (i) a set of full ρ_wave-measure on which the truncated dynamics have a uniform local existence time, (ii) a Borel–Cantelli argument to pass to a global flow, and (iii) invariance of the limiting measure. None of these is demonstrated for the specific damped wave system with exponential nonlinearity, and the sign-definite structure used earlier in Section 6 does not by itself provide the needed uniform control of escape times. The authors should either supply the full argument or explicitly reformulate Theorem 1.9 as conditional on it.","section":"§6.2, paragraph after (6.28)"}],"minor_comments":[{"comment":"In the definition of Θ_{N,M} and R_{N,M} around (5.28)–(5.29), the projector P_N appears where the text has introduced a second truncation parameter M in (5.27); this should presumably be P_M for consistency.","section":"§5.2, Eqs. (5.27)–(5.29)"},{"comment":"There are numerous typographical artifacts in the extracted text (e.g., 'equ ation', 'well-pose dness', 'approx imation'), and the notation H^s(T2) for the product space is defined only in §2.1; a unified notation table would help.","section":"Throughout"},{"comment":"The proof of Proposition 4.1 states that continuity in initial data 'follows from a standard argument' and uniqueness in the whole space is relegated to Remark 4.2; these are acceptable but should be made explicit, especially because the map is only shown to be a contraction on a ball of radius O(1).","section":"§4, Proposition 4.1 and Remark 4.2"}],"recommendation":"major_revision","confidential_remarks":"The main results of Theorems 1.1, 1.2, 1.6, and 1.8 appear to be in good shape; the revision should focus on Theorem 1.9. If the authors supply the missing Bourgain argument and fix the initial-data mismatch in §6.2, I would be willing to accept. I do not see grounds for rejection of the other claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper earns its length. The local well-posedness for the damped wave equation with exponential nonlinearity (Theorem 1.8) and the parabolic global/invariance results are supported by careful, detailed estimates; the X+Y decomposition is a genuine idea and the Brascamp-Lieb moment bounds are a solid contribution. But Theorem 1.9, the almost sure global well-posedness and Gibbs invariance for the wave equation, has a real gap in the written proof.\n\nWhat is new: the improvement over Garban's heat result (noise continuity, larger beta^2 range), the defocusing global heat result up to 4pi, and the wave local theory for beta^2<0.86pi which answers the Sun-Tzvetkov open problem in that regime. The structure of the proof is coherent: Da Prato-Debussche, then the rough/smooth split of the residual term, with the sign-definite part controlled by positivity of the wave kernel and the smooth part gaining two derivatives. The estimates in Sections 4 through 6.1 appear correct to me; the parameter choices are internally consistent.\n\nThe soft spot is Section 6.2. The truncated system (6.25) is stated with initial data (Q_N w0, Q_N w1), but the decomposition u_N = X_N + Y_N + Psi uses the untruncated stochastic convolution Psi with initial data (w0,w1). That forces (X_N+Y_N)(0) = (Q_N-I)w0 and a nonzero time derivative, whereas the Duhamel formulas in (6.26) define X_N,Y_N with zero initial data. The paper does not address this residual initial data. Moreover, the passage to global-in-time dynamics and invariance via Bourgain's invariant measure argument is deferred by 'Since it is standard, we omit details.' That adaptation is not automatic for this sign-definite truncated system, especially with the Q_N cutoffs, and it carries the full weight of Theorem 1.9.\n\nNeither issue looks fatal: one can redefine Psi with Q_N initial data, or absorb the linear evolution of (Q_N-I)w0 into X_N+Y_N, and Bourgain's argument has been implemented in similar wave settings (the paper cites [65, 37, 59, 15, 60]). But as written, Theorem 1.9 is conditional on repairs.\n\nBottom line: this is a strong paper for the local theories and the parabolic Gibbs program. The hyperbolic global/invariance theorem needs referee attention. I'd send it to peer review; a good referee can ask for the fix rather than a full rewrite.","headline":"A genuinely strong paper on singular stochastic PDEs with a real gap in the hyperbolic Gibbs theorem that a referee can likely repair.","tokens_in":67430,"tokens_out":2934,"would_cite":true,"duration_ms":32302,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35L71","35K15","60H15"],"pacs":[],"model":"deepseek-v4-flash","headline":"Exponential-noise heat and wave equations solved up to sharp β² thresholds","keywords":["stochastic nonlinear heat equation","stochastic damped nonlinear wave equation","exponential nonlinearity","Gaussian multiplicative chaos","Gibbs measure invariance","renormalized exponential","Liouville equation","random data well-posedness"],"falsifier":"Complete the omitted invariant-measure step for the truncated damped wave equation: either establish the uniform-in-$N$ a priori bounds and compactness needed to extract a global flow, or exhibit some $0<\\beta^2<\\beta^2_{\\rm wave}$ for which those bounds diverge as $N\\to\\infty$; the latter would disprove Theorem 1.9.","tokens_in":66370,"feed_emoji":"🎲","tokens_out":8424,"duration_ms":88654,"temperature":0.7,"pith_summary":"The paper studies two-dimensional stochastic heat and damped wave equations with the exponential nonlinearity $\\lambda\\beta e^{\\beta u}$, driven by additive space-time white noise. After renormalizing the exponential by subtracting its divergent variance, it proves local well-posedness for the heat equation up to $\\beta^2<8\\pi/(3+2\\sqrt2)$, global well-posedness in the defocusing case $\\lambda>0$ up to $\\beta^2<4\\pi$, and local well-posedness for the damped wave equation up to the smaller threshold $\\beta^2<(32-16\\sqrt3)\\pi/5$. For $\\lambda>0$, the paper constructs the renormalized Gibbs measures for both models and shows they are invariant under the dynamics, giving almost sure global well-posedness for Gibbs-distributed initial data. The hyperbolic statement, translated to deterministic nonlinear wave equations with random data, answers an open question left by earlier work on exponential wave equations. The interest is that these are the parabolic and hyperbolic stochastic quantization equations for the exponential-interaction field, where the exponential nonlinearity makes every moment of the driving noise contribute.","feed_headline":"Exponential-noise heat and wave equations solved up to sharp β² thresholds","feed_subtitle":"New well-posedness theory and invariant Gibbs measures for 2D Liouville equations with exponential nonlinearity.","key_machinery":"The load-bearing object is the renormalized exponential of the stochastic convolution, $\\Theta_N=:e^{\\beta\\Psi_N}: = e^{-\\beta^2\\sigma_N/2}e^{\\beta\\Psi_N}$, whose $L^p$ regularity is controlled by moment bounds on Gaussian multiplicative chaos. The argument splits the solution as $u = v + z + \\Psi$ (stochastic convolution plus deterministic linear evolution plus residual), reducing the equation to a fixed point driven by $\\Theta_N$; for the wave equation the residual is split again, $v=X+Y$, where $X$ gains a sign-definite structure under the essentially non-negative wave kernel and $Y$ gains two derivatives from the difference of the damped and undamped propagators. Dispersive and energy estimates, a compactness lemma, and a product lemma for positive distributions carry the proof, and the invariant-measure argument supplies the global wave dynamics.","core_discovery":"On the paper's own terms, the central discovery is that the two-dimensional Liouville equations with exponential nonlinearity admit a complete renormalized solution theory whose thresholds are set by the regularity of Gaussian multiplicative chaos: $\\beta^2_{\\rm heat}=8\\pi/(3+2\\sqrt2)$ arises by optimizing the smoothing exponent against the moment order needed for contraction, and $\\beta^2_{\\rm wave}=(32-16\\sqrt3)\\pi/5$ arises from a further split of the residual wave component into a sign-definite rough part and a smoother part. In the defocusing case the positivity of the chaos and a sign-definite structure make the nonlinearity bounded, extending heat well-posedness to $4\\pi$ and yielding invariant measures. The convergence statements are made in the strong sense of convergence in probability of the truncated solutions, including continuity in the noise, which was missing in prior local heat results.","pith_inferences":["The same even-moment route via the multilinear inequality behind the chaos bounds may transfer to other log-correlated non-polynomial nonlinearities whenever the correlation kernel decouples into two-point products; this is an extension, not asserted in the paper.","The wave-side dichotomy of a rough sign-definite part plus a smoother remainder is a template for oscillatory propagators whose kernels are only non-negative near the singularity; testing it on other dispersive equations with positive near-field kernels would be a natural next step.","The thresholds here are sharp for the method, not for the equation; a testable question is whether the $X/Y$ split can be pushed past $0.86\\pi$ with additional smoothing or different function spaces."],"forward_implications":["For $\\lambda>0$ and $0<\\beta^2<4\\pi$, the renormalized heat dynamics is globally well-posed and the renormalized Gibbs measure, mutually absolutely continuous with the massive Gaussian free field, is invariant.","For $0<\\beta^2<\\beta^2_{\\rm wave}$, the damped wave equation with Gibbs-distributed data is almost surely globally well-posed and $\\rho_{\\rm wave}$ is invariant.","Translated to deterministic initial data, the wave result gives almost sure global well-posedness for the nonlinear wave equation with $e^{\\beta u}$ and invariance of the Gibbs measure, settling the previously open two-dimensional case.","The local heat result applies to the sinh-Gordon equation, yielding local well-posedness with continuity in the noise for $0<\\beta^2<8\\pi/(3+2\\sqrt2)$ and, via the invariant-measure argument, almost sure global well-posedness with an invariant renormalized cosh-interaction Gibbs measure.","For the undamped wave equation, the small-time variance growth $t\\log N$ allows local well-posedness for every $\\beta^2>0$."],"supporting_citations":[{"why":"Supplies the multilinear inequality used to control even moments of the Gaussian multiplicative chaos.","marker":"[13]"},{"why":"Introduces the stochastic-convolution-plus-residual formulation used throughout.","marker":"[20]"},{"why":"Gives the prior local well-posedness threshold and positivity input whose heat range and noise continuity are improved.","marker":"[30]"},{"why":"Supplies the sine-Gordon renormalization and threshold analysis paralleled by the exponential model.","marker":"[42, 19]"},{"why":"Provides the Gaussian comparison inequality behind the uniform moment bounds on the chaos.","marker":"[46]"},{"why":"Provides the hyperbolic sine-Gordon framework, including positivity of the wave kernel and dispersive estimates, adapted to the damped wave equation.","marker":"[63]"},{"why":"Poses the deterministic random-data wave question that the hyperbolic result answers in the two-dimensional case.","marker":"[77]"},{"why":"Supplies the invariant-measure argument used to pass from local wave solutions to almost sure global solutions and Gibbs invariance.","marker":"[9, 10]"}],"fun_headline_variants":["Liouville heat/wave well-posedness pinned to chaos thresholds","Sharp β² bounds for 2D stochastic Liouville equations","Gaussian multiplicative chaos sets Liouville solution thresholds","Defocusing Liouville heat tames noise up to 4π"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof of the wave Gibbs-invariance theorem assumes, without demonstrating, that the standard invariant-measure argument applies to the truncated damped wave flow; if that adaptation fails, the claimed almost sure global well-posedness and invariance of $\\rho_{\\rm wave}$ do not follow from the local estimates alone.","fun_headline_variants_meta":{"raw":{"variants":["Liouville heat/wave well-posedness pinned to chaos thresholds","Sharp β² bounds for 2D stochastic Liouville equations","Gaussian multiplicative chaos sets Liouville solution thresholds","Defocusing Liouville heat tames noise up to 4π"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000147,"raw_usage":{"total_tokens":1123,"prompt_tokens":819,"completion_tokens":304,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":435,"completion_tokens_details":{"reasoning_tokens":236}},"tokens_in":435,"tokens_out":304,"duration_ms":4174,"temperature":1.0,"reasoning_tokens":236,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:58:15.906327+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Complete the omitted invariant-measure step for the truncated damped wave equation: either establish the uniform-in-$N$ a priori bounds and compactness needed to extract a global flow, or exhibit some $0<\\beta^2<\\beta^2_{\\rm wave}$ for which those bounds diverge as $N\\to\\infty$; the latter would disprove Theorem 1.9.","supporting_citations":[{"cited_title":"Brascamp, E","cited_arxiv_id":null,"evidence_quote":"Supplies the multilinear inequality used to control even moments of the Gaussian multiplicative chaos."},{"cited_title":"Garban, Dynamical Liouville , J","cited_arxiv_id":null,"evidence_quote":"Gives the prior local well-posedness threshold and positivity input whose heat range and noise continuity are improved."},{"cited_title":"Kahane, Sur le chaos multiplicatif, Ann","cited_arxiv_id":null,"evidence_quote":"Provides the Gaussian comparison inequality behind the uniform moment bounds on the chaos."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the hyperbolic sine-Gordon framework, including positivity of the wave kernel and dispersive estimates, adapted to the damped wave equation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Poses the deterministic random-data wave question that the hyperbolic result answers in the two-dimensional case."}],"review_version":1}