{"id":"7bfc3325-d068-4f22-9521-81191c35659b","arxiv_id":"2509.08413","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A scale-independent third-order WENO-Z variant, WENO3-ZES4, preserves third-order accuracy at first-order critical points using nonlinear weights with p=1.","lead":"The paper introduces a new version of a third-order WENO shock-capturing scheme that keeps third-order accuracy even when a critical point sits at arbitrary positions inside a grid cell, while using the smaller exponent p=1. The value is mostly for CFD practitioners who need robust, high-resolution low-order reconstruction on coarse grids.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The p=1 accuracy claim depends on Lemma 3.1 and Eqs. (19)-(20), whose printed proof/expansions are algebraically inconsistent; the argument is salvageable only if an independent expansion confirms Eq. (12).","rationale":"The reader's weakest assumption identifies the same load-bearing point: the proof's premise that the expanded local indicators satisfy Eq. (12). My independent expansion indicates the qualitative condition is likely satisfied, so the central accuracy claim may be repairable. The paper deserves conditional acceptance rather than rejection because the convergence table and the structure of Lemma 3.1 provide strong evidence that the scheme works, and the identified algebraic errors appear to be presentation errors rather than fundamental flaws. However, the central proof as printed is not rigorous enough to verify without correction, and the validation does not fully exercise arbitrary CP positions. Thus the reader's CONDITIONAL verdict is appropriate; no change is needed.","tokens_in":23623,"tokens_out":35341,"duration_ms":360048,"concrete_test":"Use a computer algebra system to expand Eqs. (17), (18), and (21) in powers of Δx around x_c = x_j + λΔx, with f'(x_c)=0 and f'', f''' nonzero. Verify: (i) the Δx^4 coefficients of β0* and β1* coincide exactly when C_β0=C_β1; (ii) the Δx^5 coefficients differ for all λ∈(-1,1) at C_β=2; (iii) substituting these expansions into the corrected Lemma 3.1 argument with equal C_α yields ω_k-d_k=O(Δx^2). Independently, run the 1D scalar advection test with CP initially offset at 0.25, 0.5, and 0.75 of a grid spacing, using N=40,80,160,320,640 and report L∞ convergence rates; asymptotic rates should be ≥2.9 for all offsets if the claim holds.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that WENO3-ZES4 is third-order at CP1 with p=1 rests entirely on Lemma 3.1 plus the assertion that the local indicators satisfy Eq. (12) with common Δx^4 coefficients and distinct Δx^5 coefficients. As printed, this is not verifiable: the Appendix derivation of ω_k contains an algebraic non-sequitur in the manipulation of (β_k/τ)^p and (β_k/β_l)^p, and a direct Taylor expansion of Eqs. (17)-(18) about x_c=x_j+λΔx does not reproduce the stated b_k = -4λ^3 + (2/3)λ - 2C_βλ ± 2C_β. The actual Δx^5 coefficients are, for β0*: (-λ^2 + (2/3-2C_β)λ - 2C_β) f''f''', and for β1*: (λ^3 + (2/3)λ + 2C_β(1-λ)) f''f'''. These still differ for all λ∈(-1,1) when C_β=2, so the required condition is plausibly true, but the paper does not supply the correct derivation. If the intended expansion or lemma conclusion is wrong, the O(Δx^2) weight condition fails and the 'regardless of position' CP1 accuracy claim is unsupported. The scalar convergence test also does not fully cover all offsets: at t=2 the CP initially at x=0 returns to a grid node, so half-node behavior is not directly measured in the reported L∞ rates.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a new third-order WENO-Z scheme, WENO3-ZES4, designed to retain third-order accuracy when a first-order critical point (CP1) occurs at an arbitrary position inside the stencil. The main theoretical contribution is Lemma 3.1, an 'accuracy-optimization lemma' stating that if the local smoothness indicators β_k share common lower-order Taylor coefficients and differ only at order n2, then normalized nonlinear weights with p=1 can gain an extra order of accuracy. The authors construct extended local indicators β_0^*, β_1^* (Eqs. (17)-(18)) and a global indicator τ_4 (Eq. (21)) so that at CP1 the required condition holds, and choose the free parameters C_β=2.0 and C_α=1.3 by a combination of Lemma 3.2 and numerical resolution tests. The paper reports a scalar advection convergence test showing third-order L∞ convergence, and a range of 1D and 2D Euler and Navier-Stokes validation cases arguing for good resolution and robustness.","tokens_in":24060,"tokens_out":18238,"duration_ms":176647,"significance":"If the central claim is correct, WENO3-ZES4 is a meaningful practical contribution: it is scale-independent, uses the favorable exponent p=1, and preserves third-order accuracy at CP1 for arbitrary critical-point position. The idea of upgrading the accuracy of nonlinear weights by exploiting a common leading-order part of β_k is interesting and goes beyond the standard WENO-Z analysis. The numerical validation is extensive and, where quantitative (Table 2), supports third-order convergence. The paper also ships explicit formulas and parameters, which is helpful for reproducibility. However, the proof machinery as printed is not fully trustworthy: the key expansion coefficient b_k in Eqs. (19)-(20) is incorrect, and the appendix proof of Lemma 3.1 contains garbled algebraic steps. These are localized and fixable, but they must be corrected before the central theoretical claim can be taken as established.","major_comments":[{"comment":"The printed coefficient b_k = -4λ^3 + (2/3)λ - 2Cβλ ± 2Cβ in the paragraph after Eq. (20) does not follow from a direct Taylor expansion of Eqs. (17)-(18) about x_c = x_j + λΔx. Expanding β_0^* and β_1^* with f'=0, f''≠0, f'''≠0 gives β_k^* = (λ^2 + Cβ) f''^2 Δx^4 + b_k Δx^5 + O(Δx^6), with b_0 = (-λ^3 + (2/3 - 2Cβ)λ - 2Cβ) f''f''' and b_1 = (-λ^3 + (2/3 - 2Cβ)λ + 2Cβ) f''f'''. Thus b_1 - b_0 = 4Cβ f''f''', which is nonzero for all λ∈(-1,1) provided Cβ>0. So the qualitative conclusion of Lemma 3.1 still holds, but the printed coefficient is wrong and the factor f''f''' is missing. Please correct the expansion and state explicitly that b_1−b_0 ≠ 0 for all λ∈(-1,1).","section":"§3.2.1, Eqs. (19)-(20)"},{"comment":"The proof contains an algebraic non-sequitur after multiplying numerator and denominator by (β_k/τ)^p. The numerator should be (β_k/τ)^p + C_{αk}, but the manuscript displays an expression with (τ/(τ β_k/β_k))^p, which is undefined and obscures the argument. The later application of a/(a+x) = 1 − x/a + O(x^2) is also not fully explicit: the 'small' quantity is x = −C_{αk} + Σ_l d_l C_{αl} + O(Δx^{n2−n1}), and the reader must verify that all O-terms are uniform. The proof is salvageable, and the lemma itself appears correct, but the printed derivation is not. Please rewrite the proof cleanly with all expansions explicitly justified.","section":"Appendix, proof of Lemma 3.1"}],"minor_comments":[{"comment":"The text says the CP1 initially at x=0 migrates to a half-node every four iterations, but at the final time t=2 it is again at a grid node. The second CP at x=-2+2x_c remains at a non-node at t=2, so the reported L∞ rates do test off-node behavior, but the paper should state this explicitly or report the L∞ error at an intermediate time when the x=0 CP is at a half-node.","section":"§4.2.1, Table 2"},{"comment":"Lemmas 2.1 and 2.2 are imported from Ref. [13] without proof. Since Lemma 2.1(2) motivates stencil extension, please state explicitly that the proofs are given in [13], or include them for self-containedness.","section":"§2.2, Lemmas 2.1 and 2.2"},{"comment":"The indexing b_1, b_2 in Eqs. (19)-(20) conflicts with the later b_k notation. Use b_0 and b_1 consistently for β_0^* and β_1^*.","section":"§3.2.1, Eqs. (19)-(20)"},{"comment":"The notation δ_j^{(n)m} is introduced informally in §2.3 and used in Lemma 3.2. Please define it precisely, including whether m denotes order of accuracy and how the δ quantities relate to finite differences.","section":"Lemma 3.2"},{"comment":"There are several typos, e.g., 'coherently free coefficient' in §2.1, 'gird' in Fig. 10 caption, and inconsistent use of Δx vs dx in Eqs. (19)-(20). A careful proofreading pass is recommended.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The core idea is promising, and the numerical validation is broad. The incorrect expansion in §3.2.1 and the garbled Appendix proof are significant because they are the load-bearing part of the accuracy claim, even though the corrected expansion still supports the conclusion. I recommend major revision, not rejection, because the errors appear fixable within the manuscript's scope. The choice of C_β and C_α is empirical; the paper should perhaps more clearly separate the mathematically guaranteed part of the scheme from the parameter-tuning component."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a legitimate incremental contribution. The central mechanism—using local smoothness indicators that share the same O(Δx^4) coefficient at a critical point but differ at O(Δx^5), so that p=1 suffices—is new and the theory behind it works. The paper deserves a serious referee, but the authors need to clean up the algebra and address the tuning criticism before the resolution claims can be taken at face value.\n\nWhat's genuinely new is Lemma 3.1 and the construction that realizes it: the β_k^* in Eqs. (17)-(18) with τ4 in Eq. (21). The idea is sound. I re-derived the Taylor expansions; the printed Δ^5 coefficients are wrong (they have -4λ^3; the correct term is -λ^3 for both), and the stress-test's own coefficients are also off. But the crucial point is that the two coefficients differ by 4Cβ for every λ, so Eq. (12) is satisfied with n1=4, n2=5 and distinct b_k. The appendix proof has garbled algebra, but the ratio argument is recoverable. These are typos to fix, not a fundamental issue.\n\nThe validation battery is extensive and the scheme completes the Mach-2000 jet and the viscous double cone, which says something real for robustness. The resolution claim is weaker than it looks, though: Cβ and Cα were tuned on the Shu-Osher problem on 240 points and the headline 'resolves the second peak' is the same test. No code or data is provided, so the plots can't be independently verified. The scalar convergence test in Table 2 only reports t=2, when the critical point has returned to a grid node; a test at a half-node time would directly support the 'regardless of position' claim.\n\nBottom line: the central accuracy claim is likely correct. The paper is for people working on WENO variants or using low-order shock-capturing schemes. With the algebra corrected and the tuning discussed honestly, it's publishable. I'd send it to referees.","headline":"A solid incremental WENO3-Z variant whose central accuracy lemma holds up; the printed algebra needs fixing and the resolution claims carry a tuning caveat.","tokens_in":24554,"tokens_out":24334,"would_cite":true,"duration_ms":202027,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["65M06","65M12"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that WENO3-ZES4, a scale-independent third-order WENO-Z variant, recovers full third-order accuracy at first-order critical points (CP1) wherever they fall inside the stencil, while using the exponent p=1 in its nonlinear w","keywords":["WENO-Z scheme","critical point accuracy","smoothness indicator","order preservation","scale independence","nonlinear weights","third-order scheme","resolution optimization"],"falsifier":"Run the paper's scalar advection test (initial condition Eq. (22), CFL 0.25, L-infinity error at t=2) with the moving first-order critical point forced to sit exactly at a half-node (lambda=1/2) on final refinement levels; the claim predicts a clean rate of 3.000. A stronger check: repeat with the critical point at other lambda values (e.g., 1/3, 0.75) and directly measure omega_k - d_k at the critical point; if the O(Delta x^2) scaling fails for some lambda, or if letting f''(x_c) approach zero drops the L-infinity rate below 3, the uniformly-at-third-order claim fails.","tokens_in":23557,"feed_emoji":"🧮","tokens_out":11735,"duration_ms":105931,"temperature":0.7,"pith_summary":"The paper sets out to remove a known weakness of the standard third-order WENO-Z shock-capturing scheme: when the solution has a first-order critical point (derivative zero, second and third derivatives nonzero), the scheme loses its third-order accuracy, and earlier fixes only work when the critical point lands on a grid node, are scale-dependent, and use large exponents that blur resolution. The core move is an accuracy-optimization lemma: if the two local smoothness indicators have identical leading error terms and differ only at the next order, the nonlinear weights gain an extra order of accuracy, enough to satisfy the optimal-order condition with exponent p=1. Using that lemma, the paper constructs new local and global smoothness indicators on a five-point stencil and fixes the free coefficients by balancing resolution against robustness, producing WENO3-ZES4. If right, the scheme holds third-order L-infinity convergence at a first-order critical point no matter where it sits in the interval, remains scale-independent, and resolves fine features that typical third-order schemes smear, such as the second density peak of the Shu-Osher problem on just 240 points.","feed_headline":"New WENO3-ZES4 holds third order wherever a critical point sits","feed_subtitle":"Earlier fixes lost order when the critical point fell between grid nodes; this one holds third order with p=1.","key_machinery":"Lemma 3.1 (the accuracy-optimization lemma): for alpha_k = d_k(1 + C_alpha (tau/beta_k)^p), if the local smoothness indicators share identical coefficients on all error terms up to order Delta x^(n2-1) and differ only in the Delta x^n2 term, and C_alpha is common across stencils, then omega_k - d_k = O(Delta x^(p(m-n1)) times Delta x^(n2-n1)); the second factor is what permits p=1. The construction satisfying its hypotheses is the pair of local indicators beta_0*, beta_1* of Eqs. (17)-(18) on the five-point stencil {x_(j-2),...,x_(j+2)}, each combining a two-point first-derivative difference with a three-point second-derivative difference, with equal coefficients C_beta0 = C_beta1 = 2.0 so t","core_discovery":"The paper claims that the accuracy of WENO-type nonlinear weights can be upgraded without raising the exponent p. Lemma 3.1 states that when local smoothness indicators expand as beta_k = sum a_l' Delta x^l' + b_k Delta x^n2 + O(Delta x^(n2+1)) with common coefficients a_l' and differing b_k, and when the constants C_alpha in all non-normalized weights are equal, then omega_k - d_k = O(Delta x^(p(m-n1))) times O(Delta x^(n2-n1)); the second factor is a bonus order. Applied to third-order WENO-Z: with the five-point local indicators of Eqs. (17)-(18) set to equal C_beta = 2.0, beta_k* behave as (lambda^2 + C_beta) f''^2 Delta x^4 + b_k Delta x^5 + ..., so n1=4, n2=5 and the Delta x^4 coeffici","pith_inferences":["Nothing in Lemma 3.1 is specific to third order; the same error-coefficient matching could lower the exponent for fifth-order WENO-Z at critical points, trading the usual p=2 for p=1 and sharper resolution, a direction the paper leaves untested.","Because the accuracy gain in Lemma 3.1 holds for any common value of C_beta, the third-order claim is insensitive to the Shu-Osher-calibrated parameters C_beta=2.0 and C_alpha=1.3; only resolution and robustness depend on them, so order tests away from the calibration point should still show the optimal rate.","The paper notes that b_k vanishes at isolated lambda values, which should produce local super-convergence; a systematic order-versus-lambda scan could locate the true worst case and possibly reveal simpler indicators satisfying Eq. (12).","For the Euler and Navier-Stokes tests the scalar proof carries over only through the local characteristic decomposition used in practice; the 'any position' guarantee for systems and for flows with discontinuities is empirical, not a proven extension."],"forward_implications":["Third-order WENO schemes can hold full accuracy at first-order critical points with exponent p=1, removing the resolution loss that large p caused in earlier fixes; the Shu-Osher second peak and valley are resolved with only 240 points.","Lemma 3.1 gives a transferable recipe: matching the lower-order error coefficients of local smoothness indicators buys n2-n1 extra orders of accuracy for the weights, relaxing how accurate the global indicator tau must be.","Scale-independence is kept: unlike WENO-NP3/-F3/-NN3/-PZ3, the new weights are invariant under rescaling of the variables, so the same computation gives the same answer at any scale.","Using a tau of only O(Delta x^5) at CP1 avoids the robustness problems that the higher-order indicators of WENO3-ZM and WENO3-ZES caused; the scheme completed every problem in its test battery, including the Mach-2000 jet and viscous double-cone flows.","The optimal-order condition is met for every lambda in (-1,1), so third-order L-infinity convergence holds whether the critical point is on a node, a half-node, or anywhere else in the interval."],"supporting_citations":[{"why":"Establishes the WENO-Z weight formula alpha_k = d_k(1 + C_alpha (tau/(beta_k + epsilon))^p) that the present scheme refines.","marker":"[4]"},{"why":"Supplies the arbitrary-critical-point-location (ACPL) analysis and Lemma 2.1 showing a three-point stencil cannot yield a scale-independent third-order scheme at CP1, forcing the stencil extension used here.","marker":"[13]"},{"why":"States the sufficient condition omega_k - d_k = O(Delta x^r) for optimal order that defines the paper's accuracy target.","marker":"[3]"},{"why":"Provides the predecessor WENO3-ZES2/ZES3 with C_beta-modified local indicators and the tau4 global indicator; its p=2 constraint motivates the p=1 goal.","marker":"[15]"},{"why":"Analyzes how the exponent p and the small parameter epsilon affect accuracy at critical points, framing the resolution cost of large p.","marker":"[6]"},{"why":"Defines the WENO-JS smoothness indicators beta_k^(r) and the parameter epsilon that the new local indicators generalize.","marker":"[2]"},{"why":"Documents that WENO3-Z attains only O(Delta x) accuracy for the difference between linear and nonlinear weights in smooth regions, setting the accuracy issue this paper addresses.","marker":"[8]"}],"fun_headline_variants":["WENO3-ZES4 keeps third order at any critical point","Critical fix: WENO3-ZES4 holds order with p=1 everywhere","Third-order WENO now critical-point-position safe","New WENO3-ZES4: order preserved at any critical point","WENO3-ZES4: third order even when CP1 sits anywhere"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The accuracy proof assumes the solution is smooth around the critical point and that both f''(x_c) and f'''(x_c) are nonzero, so the Taylor expansions of the two local indicators share the same Delta x^4 coefficient; if a discontinuity, a near-zero second derivative, or a critical point outside the five-point stencil enters, the p=1 accuracy upgrade is not established.","fun_headline_variants_meta":{"raw":{"variants":["WENO3-ZES4 keeps third order at any critical point","Critical fix: WENO3-ZES4 holds order with p=1 everywhere","Third-order WENO now critical-point-position safe","New WENO3-ZES4: order preserved at any critical point","WENO3-ZES4: third order even when CP1 sits anywhere"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000182,"raw_usage":{"total_tokens":1232,"prompt_tokens":916,"completion_tokens":316,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":660,"completion_tokens_details":{"reasoning_tokens":232}},"tokens_in":660,"tokens_out":316,"duration_ms":4701,"temperature":1.0,"reasoning_tokens":232,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T20:39:46.770300+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the paper's scalar advection test (initial condition Eq. (22), CFL 0.25, L-infinity error at t=2) with the moving first-order critical point forced to sit exactly at a half-node (lambda=1/2) on final refinement levels; the claim predicts a clean rate of 3.000. A stronger check: repeat with the critical point at other lambda values (e.g., 1/3, 0.75) and directly measure omega_k - d_k at the critical point; if the O(Delta x^2) scaling fails for some lambda, or if letting f''(x_c) approach zero drops the L-infinity rate below 3, the uniformly-at-third-order claim fails.","supporting_citations":[],"review_version":1}