{"id":"f768a34d-2d6a-4580-ae57-7218fc7c93c8","arxiv_id":"2501.04618","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A fully discrete augmented SAV scheme for the stochastic Allen-Cahn equation with multiplicative noise achieves optimal strong convergence order, with squared L2 and H1 errors bounded by C_delta(tau^{1-delta} + h^2).","lead":"This paper proves that a fast, linear numerical method for the stochastic Allen-Cahn equation with multiplicative noise converges with the same optimal accuracy as a costly fully implicit method. The result is practically relevant because stochastic phase-field simulations need many random samples, so cheaper linear schemes matter.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.2 is the load-bearing pillar of Theorem 3.1, yet its uniform discrete regularity bounds are imported from a Neumann-boundary paper with only an asserted carry-over to periodic boundaries. The rest of the proof is credible, so the verdict should hinge on verifying that transfer.","rationale":"The proof of Theorem 3.1 is a standard but intricate energy-error estimate. I checked the key displayed estimates: the decomposition of R3, the treatment of the interpolation error I6 via elementwise estimates, the Itô-isometry bound for R6, and the absorption leading to (5.25) are all plausible, and the improved R2 estimate in Lemma 4.2 is consistent with the Gagliardo-Nirenberg/Young argument. My concern is not with these steps but with their input: Lemma 4.2 supplies essentially all high-moment discrete regularity, and it is not proven in this paper. The sentence asserting that the Neumann results of [47] carry over to periodic boundary conditions is the single point on which the theorem depends and the single point not backed by an argument in the manuscript. Since the paper's own proof of (4.4b) is conditional on (4.4a), verifying the transfer is a necessary condition for Theorem 3.1 as stated. I therefore recommend conditional acceptance pending a proof (or a convincing numerical verification) of Lemma 4.2 under periodic boundary conditions; this matches the reader's identified weakest assumption and does not undermine the rest of the analysis.","tokens_in":19437,"tokens_out":31054,"duration_ms":290984,"concrete_test":"Independently re-derive [47, Lemmas 6.1-6.3] on the periodic finite-element space, tracking every boundary term in the discrete Laplacian, mass-lumped inner products, and the proof of the lag-l difference bound; verify in particular that τ^{-p}E[(∑_n τ‖Ξ^n_h‖²_{L2})^p]≤C and E[∑_m τ‖ϕ^{m+1}_h−ϕ^m_h‖^{2p}_{L2}]≤Cτ^p hold with constants independent of τ and h. A cheaper numerical smoke-test is to run the scheme for the paper's d=2 setup (ε=1, T=1, F=(φ²−1)²/4+10⁻⁵) on h=2⁻⁵,…,2⁻⁸ with τ∼h² and monitor whether the six empirical quantities in (4.4a) stay uniformly bounded as h→0; this can flag, although not prove, a failure of the carry-over.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is the strong rate estimate in Theorem 3.1. Its proof is a coherent energy argument in Section 5, and the new part of Lemma 4.2—the improved R2 estimate for the polynomial double-well potential—checks out. The genuinely load-bearing premise is Lemma 4.2 as a whole, in particular the uniform discrete regularity bounds (4.4a) and the auxiliary-variable error bound (4.4b). These bounds are used at every absorption step in Section 5: the H1-difference sums in R3 and R4, the τ^{-p}E[(∑τ‖Ξ^n_h‖²)^p] term in (4.4a), the Δh-regularity in R2 and I6, and (4.4b) for the r^n−√E_h error in R4. Lemma 4.2 is not proved in this paper: the text states only that results established in [47] for homogeneous Neumann boundary conditions 'carry over to the periodic case,' and (4.4b) is proved modulo (4.4a). If any of these uniform bounds fails—for instance if the periodic finite-element space requires different treatment of boundary terms in the discrete Laplacian or in the mass-lumped L2 estimates, or if the l=1 lag-difference bound in (4.4a) does not survive the boundary-condition change—then the absorption argument producing (5.25) does not close and Theorem 3.1 is unsupported. This is not an internal inconsistency in the displayed estimates; it is an external dependency whose transfer is asserted rather than demonstrated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes a fully discrete linear finite element scheme based on an augmented scalar auxiliary variable (SAV) method for the stochastic Allen–Cahn equation with multiplicative noise on a periodic domain. The main result, Theorem 3.1, states that under assumptions (S), (T), (C), (I), (W1), (W2), and (Z), for every δ > 0 and all sufficiently small τ, max_M E[||u(t_M)-φ^M_h||²_{L²}] + τ Σ E[||∇u(t_n)-∇φ^n_h||²_{L²}] ≤ C_δ (τ^{1-δ}+h²). The proof combines continuous regularity (Lemma 4.1) and discrete regularity/auxiliary-variable estimates (Lemma 4.2) with an energy-error argument that estimates six remainder terms in Section 5. The numerical section reports experimental orders of convergence consistent with the theorem.","tokens_in":19746,"tokens_out":16075,"duration_ms":155572,"significance":"If the proof is fully substantiated, the result is significant: it shows that a linear, unconditionally stable augmented SAV scheme matches the optimal strong convergence rates τ^{1/2-δ}+h² previously established for nonlinear implicit schemes. The proof is a careful and detailed energy-error estimate with explicit treatment of the nonlinearity, the SAV error, the augmentation terms, and the stochastic term; the improved estimate for the polynomial double-well potential in Lemma 4.2 is a concrete contribution. Numerical experiments support the theory. The main caveat is that the load-bearing discrete regularity estimates are imported from a companion Neumann-boundary paper, so the periodic boundary-condition transfer needs to be justified.","major_comments":[{"comment":"Lemma 4.2 is the load-bearing pillar of Theorem 3.1: the uniform bounds (4.4a) and the auxiliary-variable bound (4.4b) are used in the estimates of R2 (5.7), R3 (5.15), R4 (5.16), R5 (5.17), and R6 (5.24), and they are essential for the absorption argument that produces (5.25). The manuscript does not prove these bounds; it states that the results established in [47] for homogeneous Neumann boundary conditions 'carry over to the periodic case' (Section 4, paragraph before Lemma 4.2). This transfer is not demonstrated. The periodic setting changes the function-space and boundary-term structure in the discrete Laplacian (2.5), the mass-lumped estimates (2.2), and the lag-difference bounds in (4.4a), so the carry-over is not a formality. Please include a proof of (4.4a)-(4.4b) for the periodic spaces, or at least a rigorous reduction to the Neumann results that specifies exactly which boundary terms vanish and why no new terms appear.","section":"Section 4, Lemma 4.2"},{"comment":"The proof of (4.4b) is only completed for the term R2 in (4.8)-(4.9); for j∈{1,3,4,5,6} the paper refers to [47] for estimates of the form E[(Σ|Rj|)^p] ≤ Cτ^{p/2}. Because (4.4b) is applied in Section 5 with arbitrary p and δ (see (5.16)), the reader needs to know that these estimates hold under the current polynomial potential and with constants independent of the Neumann-to-periodic transfer. Please identify the exact lemmas in [47] used for each Rj or reproduce the short arguments.","section":"Section 4, Eq. (4.4b)"}],"minor_comments":[{"comment":"The statement 'for all s,t∈[0,t]' should read 'for all s,t∈[0,T]'.","section":"Section 4, Lemma 4.1"},{"comment":"After adding and subtracting P_U_h e_h^{n-1}, the first stochastic integral is a martingale difference with zero expectation; this should be stated explicitly, since the displayed inequality alone does not show how that term is removed.","section":"Section 5, Eq. (5.18)"},{"comment":"The absorption step is terse; please state explicitly how the (1+6α)Σ_{n=1}^M τ E||e^n_h||² term is split between the e^M term and the Gronwall sum, so that the coefficient (1/4-(1+6α)τ) is transparent.","section":"Section 5, Eq. (5.25)"},{"comment":"The notation 'eh = 2^{-8}' appears to be a typo; it should probably read 'ĥ = 2^{-8}'.","section":"Section 6, Table 1"}],"recommendation":"major_revision","confidential_remarks":"The main technical gap is the unproved Neumann-to-periodic transfer of Lemma 4.2. The rest of the proof is coherent, and I do not see an internal contradiction in the displayed estimates. The paper relies heavily on the author's companion papers [46,47]; that is legitimate, but it makes it important that the periodic transfer be supplied before publication. I would not insist on reproducing every estimate from [47], but the boundary-condition change should be explicitly addressed. The numerical section is short but adequate for the stated purpose."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuine result, not a rehash. Metzger proves strong convergence rate tau^{1-delta}+h^2 for the augmented SAV scheme for stochastic Allen-Cahn with multiplicative noise. The scheme was introduced in his earlier IMA paper [47], which only got convergence along subsequences; the rate theorem is new, and it matches the optimal rates Majee and Prohl proved for a nonlinear implicit scheme. The proof is a careful energy argument. The new part I can verify — the improved remainder estimate for polynomial double-well potentials — checks out. The numerical section, while short, gives EOCs consistent with the theory.\n\nThe soft spots are real but not fatal. Lemma 4.2 is load-bearing: the discrete regularity bounds in (4.4a) and the auxiliary-variable error (4.4b) are used at every absorption step in Section 5. Those bounds are not proved here; they're imported from [47], which was written for homogeneous Neumann data. The paper says the results 'carry over to the periodic case' without showing the discrete boundary terms behave. For periodic FE with mass lumping that is plausible, but I would want a referee to check it explicitly. If the transfer fails, the proof doesn't close.\n\nAlso worth flagging: the numerical experiments use rho(phi) = max(1-phi^2,0), which is not C^2 — it violates assumption (C) by a mile. That doesn't make the theorem wrong, but it means the numerics are outside the stated assumptions. Either replace rho in the test or relax (C). And there's no code or data shipped, so the rate table is not independently reproducible; that's minor but worth mentioning.\n\nThe citation pattern is fine: heavy self-citation, but the self-cited papers actually contain the imported estimates and the scheme, so it's not gratuitous.\n\nBottom line: central theorem is likely correct and is a useful step — linear, unconditionally stable, optimal rates for Monte Carlo. I'd send it to a serious referee, asking specifically for a careful check of Lemma 4.2 and the periodic-carry-over. The author should also fix the numerical/assumption mismatch. Not a takedown; it's a solid paper that needs minor to moderate tightening.","headline":"Solid new rate theorem for a linear SAV scheme, with a load-bearing carry-over claim that a referee should check; numerics use a coefficient outside the stated assumptions.","tokens_in":20257,"tokens_out":2781,"would_cite":true,"duration_ms":28148,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H35","65M60","60H15","65M12"],"pacs":[],"model":"deepseek-v4-flash","headline":"A linear, unconditionally stable SAV scheme attains the optimal strong rates for the stochastic Allen–Cahn equation with multiplicative noise.","keywords":["stochastic Allen-Cahn equation","multiplicative noise","finite elements","strong rate of convergence","scalar auxiliary variable method","periodic boundary conditions","unconditional stability"],"falsifier":"Compute, on the periodic domain and with the finite-dimensional noise of Section 6, the experimental order of convergence in $\\tau$ at fixed small $h$ (scaling $\\tau\\sim h^2$); if the temporal EOC stabilizes below $1/2-\\delta$ for some fixed $\\delta>0$, Theorem 3.1 is false. A more direct falsifier is to test the imported Lemma 4.2: check numerically that $E[\\max_{0\\le m\\le N}|r_h^m-\\sqrt{E_h(\\varphi_h^m)}|^p]\\le C\\tau^{p(1/2-\\delta/2)}$ and that the bounds (4.4a) hold for periodic boundary conditions; since the proof assumes rather than proves this transfer, failure of either bound would collapse the argument.","tokens_in":19214,"feed_emoji":"🧮","tokens_out":19533,"duration_ms":160714,"temperature":0.7,"pith_summary":"The paper proves that the augmented scalar auxiliary variable (SAV) method—a fully discrete, linear finite element scheme—approximates solutions of the stochastic Allen–Cahn equation with multiplicative noise at optimal strong rates: for every $\\delta>0$, the expected $L^2(O)$ error at the final time together with the accumulated $H^1(O)$ error is bounded by $C_\\delta(\\tau^{1-\\delta}+h^2)$. This matches the rates previously established for a nonlinear, implicit scheme, but the SAV scheme solves only linear systems at each step and is unconditionally stable. The essential device is a modification of the SAV auxiliary-variable update that compensates for the poor temporal regularity of stochastic solutions, so the auxiliary variable tracks $\\sqrt{E_h(\\varphi_h^n)}$ to order $\\tau^{1/2-\\delta/2}$. The result makes linear, energy-stable SAV-type discretizations provably reliable alternatives to implicit methods for phase-field SPDEs.","feed_headline":"Linear SAV scheme matches optimal rates for stochastic Allen–Cahn","feed_subtitle":"The augmented SAV method matches the accuracy of nonlinear implicit schemes while solving only linear systems.","key_machinery":"The key mechanism is the augmented SAV update. The auxiliary variable $r_h^n$ is meant to approximate $\\sqrt{E_h(\\varphi_h^n)}$, where $E_h(\\zeta):=\\int_{\\mathcal{O}} I_h\\{F(\\zeta)\\}\\,\\mathrm{d}x$ and $F(\\varphi)=\\tfrac14(\\varphi^2-1)^2+\\gamma$ is the shifted double-well potential with $\\gamma>0$. In a standard SAV scheme the update of $r_h^n$ is a first-order Taylor approximation of $\\sqrt{E_h(\\varphi_h^n)}$, leaving an error of order $|\\varphi(t_n)-\\varphi(t_{n-1})|^2$ per step; Brownian solutions are only Hölder continuous with exponent below $1/2$, so that error does not vanish fast enough. The augmentation adds two terms that linearly reproduce the second-order Taylor terms involving $\\Phi_h(\\varphi_h^{n-1})\\Delta_n\\xi^\\tau$, where $\\Delta_n\\xi^\\tau:=W(t_n)-W(t_{n-1})$, lowering the remaining auxiliary-variable error to $\\tau^{1/2-\\delta/2}$ (Lemma 4.2(4.4b)). That bound, together with the uniform discrete regularity estimates (4.4a), is what allows the error analysis to reach $\\tau^{1-\\delta}+h^2$.","core_discovery":"The central claim (Theorem 3.1) is that, under assumptions (S), (T), (C), (I), (W1), (W2), and (Z), for every $\\delta>0$ there is a constant $C_\\delta$ independent of $\\tau$ and $h$ such that for all sufficiently small $\\tau$, $$\\max_{0\\le M\\le N}\\mathbb{E}[\\|u(t_M)-\\varphi_h^M\\|_{$L^{2}$(O)}^2]+\\tau\\sum_{n=1}^N\\mathbb{E}[\\|\\nabla u(t_n)-\\nabla\\varphi_h^n\\|_{$L^{2}$(O)}^2]\\le C_\\delta(\\$tau^{{1-\\delta}}$+$h^{2}$).$$ The estimate controls the maximal expected $L^2$ error over all time levels and the expected accumulated $H^1$ error, giving temporal order $1/2-\\delta$ and spatial order $1$. The scheme is linear in the unknowns $(\\varphi_h^n,r_h^n)$ and unconditionally stable with respect to the SAV modified energy, and the proof passes through a one-step error decomposition in which the only temporal loss comes from the auxiliary-variable mismatch at order $\\tau^{1/2-\\delta/2}$.","pith_inferences":["The same augmentation mechanism should transfer to other gradient-flow SPDEs whose solutions have Hölder regularity below $1/2$; in particular, this rate analysis provides a template for upgrading the convergence-along-sequences result for stochastic Cahn–Hilliard equations with dynamic boundary conditions to genuine strong rates.","A sharper bound in Lemma 4.2(4.4b) would plausibly remove the $\\delta$-loss from the final temporal rate, since the paper's error decomposition isolates the auxiliary-variable mismatch as the source of the $\\tau^{1-\\delta}$ factor; this upgrade is not claimed in the paper.","A practical diagnostic suggested by the proof is to monitor $|r_h^n-\\sqrt{E_h(\\varphi_h^n)}|$ along sample paths: the theory predicts this quantity should exhibit the same experimental order of convergence as the phase-field error."],"forward_implications":["A linear, unconditionally stable scheme now provably attains the same strong convergence order as the nonlinear implicit scheme of [43] for the stochastic Allen–Cahn equation.","The error bound suggests the practical scaling $\\tau\\sim h^2$, which is exactly the scaling used in the paper's numerical experiments; at this scaling the observed temporal order is close to $1/2$ even on the coarsest meshes tested.","The auxiliary-variable error is controlled at rate $\\tau^{1/2-\\delta/2}$, the specific property that lets the SAV update survive the low Hölder regularity of stochastic solutions.","For multiplicative noise of the type considered here, the paper's numerical experiments indicate that the augmented SAV scheme reproduces the accuracy of the implicit scheme of [43] while requiring about half the computational time."],"supporting_citations":[{"why":"Introduces the augmented SAV scheme and supplies the discrete existence, uniqueness, and uniform regularity bounds (4.4a) that Lemma 4.2 imports, together with the auxiliary-variable error estimate that the paper sharpens for the polynomial double-well potential.","marker":"[47]"},{"why":"Establishes the optimal strong rates $\\tau^{1/2-\\delta}+h^2$ for a nonlinear implicit scheme; these are the rates that Theorem 3.1 matches.","marker":"[43]"},{"why":"Provides existence, uniqueness, and the continuous regularity estimates (Lemma 4.1) for solutions of the stochastic Allen–Cahn equation used as input by the error analysis.","marker":"[11]"},{"why":"Supplies the discrete Gronwall inequality used in the final absorption step that turns the one-step estimates into the global bound of Theorem 3.1.","marker":"[53]"},{"why":"Provides the finite element interpolation error estimates (Lemma 2.1) used to control mass-lumping and nonlinearity interpolation terms in the error decomposition.","marker":"[44]"},{"why":"Provides the $L^2$-projection stability and error estimates (2.4) used to separate the spatial discretization error from the temporal SAV error.","marker":"[19]"},{"why":"Supplies the discrete Gagliardo–Nirenberg inequality used in the proof of Lemma 4.2 and in estimating the nonlinear consistency terms.","marker":"[23]"},{"why":"Gives the standard piecewise-linear interpolation error bound used element-wise in the estimate of the nonlinear consistency term $I_6$.","marker":"[12]"}],"fun_headline_variants":["Augmented SAV hits optimal strong rates for stochastic Allen–Cahn","Linear scheme matches nonlinear accuracy for stochastic Allen–Cahn","Optimal error bounds for stochastic Allen–Cahn via linear SAV","SAV scheme overcomes low regularity in stochastic Allen–Cahn","Strong convergence of linear SAV for stochastic Allen–Cahn"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is Lemma 4.2: the uniform discrete regularity bounds (4.4a) and the auxiliary-variable error bound (4.4b), which are imported from the earlier analysis for homogeneous Neumann boundary conditions and asserted, without proof, to remain valid for the periodic boundary conditions of this paper, together with the improved remainder estimate for the polynomial double-well potential.","fun_headline_variants_meta":{"raw":{"variants":["Augmented SAV hits optimal strong rates for stochastic Allen–Cahn","Linear scheme matches nonlinear accuracy for stochastic Allen–Cahn","Optimal error bounds for stochastic Allen–Cahn via linear SAV","SAV scheme overcomes low regularity in stochastic Allen–Cahn","Strong convergence of linear SAV for stochastic Allen–Cahn"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000638,"raw_usage":{"total_tokens":2949,"prompt_tokens":968,"completion_tokens":1981,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":584,"completion_tokens_details":{"reasoning_tokens":1891}},"tokens_in":584,"tokens_out":1981,"duration_ms":14608,"temperature":1.0,"reasoning_tokens":1891,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:28:49.187585+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute, on the periodic domain and with the finite-dimensional noise of Section 6, the experimental order of convergence in $\\tau$ at fixed small $h$ (scaling $\\tau\\sim h^2$); if the temporal EOC stabilizes below $1/2-\\delta$ for some fixed $\\delta>0$, Theorem 3.1 is false. A more direct falsifier is to test the imported Lemma 4.2: check numerically that $E[\\max_{0\\le m\\le N}|r_h^m-\\sqrt{E_h(\\varphi_h^m)}|^p]\\le C\\tau^{p(1/2-\\delta/2)}$ and that the bounds (4.4a) hold for periodic boundary conditions; since the proof assumes rather than proves this transfer, failure of either bound would collapse the argument.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the augmented SAV scheme and supplies the discrete existence, uniqueness, and uniform regularity bounds (4.4a) that Lemma 4.2 imports, together with the auxiliary-variable error estimate that the paper sharpens for the polynomial double-well potential."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the optimal strong rates $\\tau^{1/2-\\delta}+h^2$ for a nonlinear implicit scheme; these are the rates that Theorem 3.1 matches."},{"cited_title":"Breit and A","cited_arxiv_id":null,"evidence_quote":"Provides existence, uniqueness, and the continuous regularity estimates (Lemma 4.1) for solutions of the stochastic Allen–Cahn equation used as input by the error analysis."},{"cited_title":"Thomée,Galerkin finite element methods for parabolic problems, Springer Series in Computational Mathematics, vol","cited_arxiv_id":null,"evidence_quote":"Supplies the discrete Gronwall inequality used in the final absorption step that turns the one-step estimates into the global bound of Theorem 3.1."},{"cited_title":"Metzger,A convergent finite element scheme for a fourth-order liquid crystal model, IMA J","cited_arxiv_id":null,"evidence_quote":"Provides the finite element interpolation error estimates (Lemma 2.1) used to control mass-lumping and nonlinearity interpolation terms in the error decomposition."},{"cited_title":"Ern and J.-L","cited_arxiv_id":null,"evidence_quote":"Provides the $L^2$-projection stability and error estimates (2.4) used to separate the spatial discretization error from the temporal SAV error."},{"cited_title":"Grün, F Guillén-González, and S","cited_arxiv_id":null,"evidence_quote":"Supplies the discrete Gagliardo–Nirenberg inequality used in the proof of Lemma 4.2 and in estimating the nonlinear consistency terms."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the standard piecewise-linear interpolation error bound used element-wise in the estimate of the nonlinear consistency term $I_6$."}],"review_version":1}