{"id":"a13f04a3-d47d-4802-b65d-b7a4db35f7b6","arxiv_id":"2507.09278","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A semi-discrete and a fully discrete finite difference scheme for a reaction-diffusion PDE-ODE system with a Hölder stochastic boundary condition converge to the unique continuum solution, with the fully discrete theorem containing a residual gap.","lead":"This paper proves that a space-discrete finite difference scheme for a nonlinear reaction-diffusion system with a randomly fluctuating boundary condition converges to the true continuum random solution as the grid is refined. It provides the first convergence proof for this marble-sulphation type model with a stochastic dynamical boundary, using Besov-space compactness and a splitting of the solution into a random heat part and a deterministic-boundary nonlinear part.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 43's uniform bound is invoked self-referentially: applying Proposition 42 with f=s requires the very K-bound (65) that Proposition 43 is supposed to establish, and the needed bootstrap from Theorem 46 is not written.","rationale":"The reader's weakest_assumption correctly identifies the circular closure of Proposition 43: the nonlinear estimate is derived by substituting f=s into Proposition 42, but Proposition 42 is stated under conditions (52) that include exactly the norm bound Proposition 43 is supposed to prove. This is not a merely cosmetic issue: Theorem 56's compactness argument requires the uniform-in-h bound (65), and Proposition 43 as written does not supply it. The paper provides substantial independent structure — explicit semigroup representations, Feynman-Kac formulas, and detailed Besov estimates — but none of these replaces the missing bootstrap. The concern is repairable, since the linearized estimate (63) has an absorption structure that could yield L ≤ 2μ, and Theorem 46's local existence plus a continuity argument might close the gap. However, that argument is not present, so the conditional verdict is appropriate. I agree with the reader's identification and do not see a more load-bearing concern elsewhere: Theorem 63 is indeed an overclaim for L^∞H^1 convergence, but it is secondary to the central semi-discrete convergence claim, whereas Proposition 43 directly underpins Theorem 56. No formal verification or reproduced code is provided, so this analytical gap is the decisive issue.","tokens_in":33500,"tokens_out":4772,"duration_ms":59551,"concrete_test":"Derive the direct energy estimate for the nonlinear system (28) without invoking (52). Multiply the first equation in (28) by s (or use the φ-weighted identity), sum over hZ_+, and bound all terms by data using Young and Gronwall. Verify whether one obtains sup_t ||s||^2_{L^2} + ∫_0^T ||D_h^+ s||^2_{L^2} ≤ C(η,λ,C_0,φ_min,B,T,||ψ||_{C^β}), independent of h. Equivalently, on the maximal existence interval, define S(t) = sup_{τ≤t} ||s(τ)||^2_{L^2} + ∫_0^t ||D_h^+ s||^2_{L^2}; using Proposition 42 on [0,t] with f=s requires S(t) ≤ K, so prove S(t) ≤ μ + (1/2)∫_0^t ||D_h^+ s||^2, hence S(t) ≤ 2μ, and then check the remaining conditions in (52) (positivity, boundary value, L∞ bound) hold for s. If this closes, the concern is resolved and only a presentation fix is needed; if it does not, Proposition 43 fails and Theorem 56 lacks its key uniform bound.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step for Theorem 56 is the uniform-in-h a priori bound (65). Its proof reads: 'By Proposition 42 one can easily derive that the same estimate is true for the nonlinear system, taking f=s and applying inequality (64).' But Proposition 42's hypotheses, conditions (52), require ||f||^2_{C([0,T],L^2)} + ||D_h^+ f||^2_{L^2 L^2} ≤ K. Taking f=s, this is precisely the left-hand side of (65), the bound Proposition 43 is meant to prove. Inequality (64) is not a free estimate: it is obtained from (63) only after choosing K ≥ 2μ under the hypothesis that f satisfies (52). Thus, as written, the proof is circular unless a separate bootstrap or continuation argument is supplied. Theorem 46 gives only local-in-time existence up to a maximal T_h and does not itself provide the uniform bound. If this bootstrap cannot be closed, the compactness argument in Theorem 56 has no uniform bound in B^k_{p,p} to work with, and the convergence conclusion is unsupported. A possible repair exists: from (63) with f=s, absorption gives L ≤ 2μ when the remaining conditions in (52) hold for f=s, but the paper does not write this step or verify those conditions. This is the most load-bearing gap because every convergence statement passes through the a priori estimates of Section 6.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies a finite-difference semidiscretization of the random reaction-diffusion-ODE system (1)-(6) on the half-line, with a Holder-continuous stochastic Dirichlet boundary condition psi of regularity beta in (1/4,1/2). The authors split s=u+v into a random discrete heat equation with boundary psi and a nonlinear nonlocal discrete system for v with zero boundary, derive Besov-type a priori estimates for u and v, and then prove that piecewise-constant extensions of the discrete solutions converge in L^p([0,T],B^k_{p,p}(R+)) to the unique continuum mild/weak solution. A fully discrete scheme is also claimed to converge. The main result is Theorem 56; the proof relies on a uniform a priori bound (Proposition 43), compact embeddings of discrete Besov spaces, and a weak formulation of the discrete system.","tokens_in":33628,"tokens_out":12610,"duration_ms":149621,"significance":"If valid, Theorem 56 would be a meaningful first convergence theorem for this class of strongly nonlinear PDE-ODE systems with stochastic dynamical boundary conditions, and the use of discrete Besov spaces is well matched to the low boundary regularity. The paper's explicit discrete heat-kernel representation, the Feynman-Kac formula in Appendix A, and the reliance on the heat-kernel regularization from [8] are concrete strengths. However, the central uniform-in-h estimate is not closed as written, and the final strong-convergence arguments for both the semidiscrete and fully discrete schemes have gaps; these issues must be repaired before the theorem can be accepted.","major_comments":[{"comment":"The proof of the uniform bound (65) is circular as written. Proposition 42 is stated for any f satisfying (52), and the third condition in (52) is exactly ||f||^2_{C([0,T],L^2(Omega_h^+))}+||D_h^+ f||^2_{L^2 L^2}<=K. Taking f=s, this is the left-hand side of (65), the estimate Proposition 43 is supposed to prove. Moreover, Theorem 46 invokes Proposition 43 to extend the local solution to time T_fin, so there is no independent global existence result to supply (52). A standard bootstrap repair would consist in first obtaining the structural conditions in (52) for the local nonlinear solution (positivity, s(.,0)=psi, ||s||_{L^infty}<=eta), then applying (63) with f=s and absorbing the term (1/2)||D_h^+ s||^2 to obtain L<=2mu; the paper does not provide these steps.","section":"§6, Proposition 43"},{"comment":"The convergence proof only establishes weak convergence of E_h s_h in L^2([0,T],H^1) (and weak convergence of its difference quotients), whereas the theorem asserts strong convergence in L^p([0,T],B^k_{p,p}(R+)) for p in [1,2] and k<1/p. No uniform bound on the time derivative partial_t s_h (for instance in L^2_t H^{-1} or in a time-Besov space) is given, so the compactness needed to upgrade the weak convergence to the claimed strong convergence is absent. Strong convergence of c_h is proved separately, but it does not imply the missing time compactness for s_h; this gap is load-bearing because the final conclusion of Theorem 56 is exactly the strong convergence in L^p_t B^k_{p,p}.","section":"§7.2, proof of Theorem 56"},{"comment":"As stated, Theorem 63 cannot hold: the fully discrete interpolant is piecewise constant in space, so it is not an H^1(R+) function, and Theorem 56 only provides convergence in L^p([0,T],B^k_{p,p}(R+)), not in L^infty([0,T],H^1(R+)). In the proof the first term ||s-s_h||_{L^infty H^1} is therefore not available. The theorem and its proof need to be reformulated in a norm compatible with the interpolant (for example L^p_t B^k_{p,p}), or the authors must introduce an H^1-conforming interpolation together with the required estimates.","section":"§7.3.2, Theorem 63"}],"minor_comments":[{"comment":"The condition 'r<2beta+p' appears to be a typo; the proof bounds u in W^{p,r} whenever r+infty}' that should be 'lim_{h->0}', and the statement opens with 'Le us suppose' instead of 'Let us suppose'.","section":"Theorem 63"},{"comment":"Lemma 30 is imported from [8] without proof; since the heat-kernel regularization estimate is used repeatedly (Lemmas 31-32, Proposition 34), stating the precise Lemma 2.10 assumption and its proof or a self-contained derivation would improve verifiability.","section":"Lemma 30"}],"recommendation":"major_revision","confidential_remarks":"The paper relies on [35] for well-posedness of the target continuum system and on [8] for the discrete heat-kernel Besov estimates; both are by overlapping authors. The editor may wish to ensure that the novel contribution relative to these works is clearly delineated, especially in Section 6 where the linearized estimates follow [35]. The fully discrete Theorem 63 as stated is not supported and will require a substantial reformulation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Key takeaway: the semi-discrete convergence result is a real and likely correct contribution, but the proof has a load-bearing bootstrap gap in Proposition 43, and the fully discrete theorem overclaims in a way that should be corrected.\n\nThe genuinely new thing here is the proof that the space-discrete scheme converges to the continuum system for a stochastic dynamical boundary with Hölder β∈(1/4,1/2) noise. [35] gives continuum well-posedness, [4] gives boundedness/stability of a scheme; neither does convergence. The splitting into u+v and the discrete Besov machinery are well matched to the problem, and most of the a priori estimates are carefully done. If the gap I mention is fixed, Theorem 56 is a plausible and useful result.\n\nThe soft spot is Proposition 43. The proof says 'taking f=s' in Proposition 42, but Proposition 42's hypothesis (52) includes exactly the norm bound that Proposition 43 is supposed to prove. That is circular as written. A standard continuation/absorption argument can repair it: from (63) with f=s one gets A ≤ μ + A/2, hence A ≤ 2μ, and then choose K ≥ 2μ. But the paper doesn't write this step, and since Proposition 45 and Theorem 56 both rely on the uniform bound (65), this gap is load-bearing. The local existence theorem (46) is given, but the bootstrap from maximal T_h to uniform bounds is not spelled out.\n\nThe other real problem is Theorem 63. It claims convergence to L∞([0,T], H^1(R+)) for a piecewise-constant interpolant, which is not in H^1, and its proof invokes Theorem 56 for a stronger norm than that theorem supplies. The fully discrete section should either be weakened to convergence in the same Besov spaces of Theorem 56 or the interpolation operator changed (e.g., piecewise linear in space). There are also mechanical issues: the parameter check in Lemma 32 has the inequality reversed (it should be 1/4 > (k−1)/2), and Theorem 63's proof contains 'h→∞' instead of 'h→0'.\n\nNone of this destroys the central semi-discrete claim, which looks sound modulo the bootstrap. This is a paper that merits serious refereeing, but the referee should insist on a written bootstrap for Proposition 43 and a corrected or weakened Theorem 63.","headline":"A real new convergence theorem for a semi-discrete scheme with stochastic dynamical boundary, but Proposition 43 has a fixable circular step and the fully discrete theorem overclaims.","tokens_in":34304,"tokens_out":4343,"would_cite":true,"duration_ms":52085,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H35","65M06","65M12"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that a space-discrete approximation of a random reaction-diffusion system with stochastic boundary condition converges to the unique continuum solution.","keywords":["random dynamical boundary","semi-discrete scheme","SPDE","convergence","Besov norms","discrete Besov spaces","reaction-diffusion system","marble sulphation"],"falsifier":"A concrete way to test the central claim is to compute $\\|s_h\\|_{L^2([0,T],H^1(\\Omega_h^+))}$ for decreasing $h$ on the fully discrete scheme with a Pearson-process boundary (8) satisfying the paper's assumptions; if it is unbounded as $h\\to0$, then Proposition 43 fails and the compactness argument of Theorem 56 has no uniform bound to work with. Conversely, a rigorous counterexample with initial data and a boundary path satisfying (7) for which this norm diverges would disprove the convergence theorem.","tokens_in":33154,"feed_emoji":"🧮","tokens_out":10303,"duration_ms":100429,"temperature":0.7,"pith_summary":"The paper proves that a space-discrete (lattice) approximation of a highly nonlinear reaction-diffusion PDE-ODE system with a stochastic dynamical boundary condition converges to the unique weak solution of the continuum model as the mesh size tends to zero. The boundary noise is only Hölder continuous with exponent $\\beta\\in(1/4,1/2)$, so the solution is too irregular for classical Sobolev convergence; the paper shows the natural convergence occurs in time-space Besov spaces $B^k_{p,p}(\\mathbb{R}_+)$, which measure fractional smoothness in both time and space, for $p\\in[1,2]$ and $k<1/p$. The proof splits the solution into a discrete heat equation absorbing the random boundary plus a deterministic-boundary nonlinear system, obtains uniform a priori bounds on both parts, and closes with compact embedding arguments. A fully discrete forward-time centered-space scheme is also shown to converge to the continuum solution when the time step goes to zero suitably. If the result holds, it provides a rigorous numerical justification for simulations of pollutant-driven marble sulphation under random environmental boundary fluctuations.","feed_headline":"Lattice scheme with random boundary noise: convergence proved","feed_subtitle":"Space-discrete solutions converge to the unique continuum solution in Besov spaces as mesh size goes to zero.","key_machinery":"The argument rests on three load-bearing pieces. First, the splitting $s=u+v$ (Proposition 21): $u$ is the solution of the discrete heat equation on the half-lattice with zero initial data and boundary value $\\psi_t$, which absorbs the irregularity of the boundary path, while $v$ solves the nonlinear nonlocal system with deterministic initial and zero boundary data, coupled to $u$. Second, the regularization estimates for the discrete heat semigroup $e^{t\\Delta_h}$ in Besov spaces (Lemmas 30-32), which transfer time regularity of $\\psi$ into spatial regularity of $u$: the solution $u$ is bounded in $L^\\infty_t W^{1,2}_x$ uniformly in $h$ (Proposition 34). Third, the linearized system (53)-(55) with a generic function $f$ in place of $s$, for which Propositions 38-42 give uniform $L^2$-type bounds; the nonlinear estimate (Proposition 43) is then obtained by formally taking $f=s$. Finally, the piecewise-constant extension operator $E_h$ (Definition 47) maps discrete Besov spaces continuously into continuum Besov spaces (Theorem 48), so the compactness argument of Theorem 56 yields convergence in $L^p([0,T],B^k_{p,p}(\\mathbb{R}_+))$.","core_discovery":"The central claim is Theorem 56: let $(s_h,c_h)$ be the unique solution of the space-discrete system (72) with initial data converging in $H^1$ to $(s_0,c_0)$ and a boundary process $\\psi$ satisfying the Hölder condition (7). Then the piecewise-constant extensions $(E_h s_h, E_h c_h)$ converge in $L^p([0,T], B^k_{p,p}(\\mathbb{R}_+))$ for every $p\\in[1,2]$ and every $k<1/p$ to the unique weak solution $(s,c)$ of the continuum system (67)-(68). The proof uses the splitting $s=u+v$ from Proposition 21, where $u$ solves the space-discrete heat equation with the random boundary condition and $v$ solves a nonlinear, nonlocal space-discrete system with deterministic boundary data; uniform-in-$h$ a priori estimates for both components are obtained through discrete heat-kernel regularization in Besov spaces and a linearization procedure, and compact embeddings of time-space Besov spaces on the lattice transfer the estimates to the continuum. The paper further proves that a fully discrete FTCS scheme converges to the space-discrete system with rate $O(k)$ for fixed $h$, and therefore to the continuum solution as $h\\to0$ under the stability conditions (73)-(74).","pith_inferences":["The circular step in Proposition 43 (taking $f=s$ against condition (52)) is a gap in the written proof: a complete argument would need an intermediate bootstrap showing the local-in-time solution of Theorem 46 exists on $[0,T]$ with a norm bound uniform in $h$ before applying the linearized estimate. This may be fixable, but as written the uniform bound that Theorem 56 relies on is not fully esta","The same splitting plus discrete-heat-kernel regularization in Besov spaces could plausibly handle other boundary-driven singular SPDEs on lattices, such as equations with interior multiplicative noise or with boundary noise of lower Hölder regularity; the paper's Lemma 32 already shows how far the heat kernel can compensate.","Remark 62 notes the $O(k)$ rate for fixed $h$ is not uniform in $h$; tracking how the Lipschitz constants of the discrete nonlinearity grow as $h\\to0$ could yield an explicit rate in $h$, which the paper does not provide.","A cleaner convergence proof might avoid the $f=s$ bootstrap entirely by proving the uniform bound directly from a maximum principle or energy estimate for the nonlinear system (28), which would also strengthen Theorem 46 from local to global well-posedness in one step."],"forward_implications":["The semi-discrete scheme (25)-(26) is a convergent numerical method: its piecewise-constant extensions converge in $L^p([0,T], B^k_{p,p}(\\mathbb{R}_+))$ to the unique continuum weak solution, giving a rigorous basis for simulations of the sulphation model.","The fully discrete FTCS scheme, with time step $k(h)\\to0$ chosen under the stability conditions (73)-(74), converges to the continuum solution in $L^\\infty([0,T],H^1(\\mathbb{R}_+))$ (Theorem 63).","The convergence holds for every boundary process with the stated Hölder regularity, including the Pearson process (8) and fractional Brownian motion with Hurst index $>1/4$, so the result is not tied to one noise model.","Time-space Besov spaces $B^k_{p,p}$ are the correct convergence spaces at this low boundary regularity: classical $L^2(0,T;H^1)$ convergence would generally fail because the solution's spatial regularity is tied to the Hölder exponent of the boundary path.","The a priori estimates are uniform in $h$ for $0