{"id":"f4a68da9-6309-429a-be2f-6e1380af41b4","arxiv_id":"2508.18220","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A multi-field, multi-state quadratic manifold framework is claimed to reduce thermo-mechanically coupled damage-plasticity simulations while avoiding the slow-convergence Kolmogorov barrier of linear reduced models.","lead":"This paper proposes a reduced-order model built on a curved quadratic manifold to speed up simulations of materials that crack, heat up, and deform together. It claims to escape the Kolmogorov barrier that makes simple linear approximations fail on damage problems, which could make large industrial multiphysics simulations much cheaper.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Monotonic error decrease is not the same as mitigating the Kolmogorov barrier; the abstract's success criterion may be insufficient to support the central claim.","rationale":"The reader's UNVERDICTED verdict is appropriate because the full text is inaccessible and verification is impossible. However, the abstract itself provides an additional, identifiable logical weakness that is independent of the corruption: the claimed success metric, 'smooth and monotonic decrease in error as the number of modes increases,' does not by itself establish mitigation of the Kolmogorov barrier. Since the n-width is inherently non-increasing in n, linear ROMs already enjoy monotone best-approximation error; the barrier is about slow decay. Thus the burden is to show a faster decay rate, not merely monotonicity. The decoupling concern raised by the reader is plausible but secondary: even if the manifold faithfully represents coupled fields, a monotone-error benchmark would not substantiate the central claim. I recommend CONDITIONAL acceptance: the paper should be accepted only if the benchmark section, once recovered, shows a quantitatively faster decay for the quadratic manifold relative to the linear subspace n-width at comparable cost. If the full text remains unavailable or the benchmark shows only monotonicity, the central claim should be treated as unverified rather than established.","tokens_in":11860,"tokens_out":3584,"duration_ms":49240,"concrete_test":"Obtain a clean copy of the benchmark section and plot the relative error against the total latent dimension n for both the linear ROM and the proposed quadratic manifold on the same thermo-mechanically coupled damage-plasticity problem. Fit log-log slopes over the asymptotic range of n. Check whether the quadratic manifold's slope is significantly steeper (more negative) than the linear ROM's n-width slope, and whether the comparison is made at matched total online dimension (accounting for the quadratic manifold's extra quadratic coefficients). If the only visible evidence is monotonic curves without a rate comparison, the 'mitigates the Kolmogorov barrier' claim is not demonstrated.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is that the multi-perspective quadratic manifold 'mitigates the Kolmogorov barrier' of linear ROMs. The evidence cited in the abstract is 'a smooth and monotonic decrease in error as the number of modes increases.' But the Kolmogorov n-width is, by definition, non-increasing in the approximation dimension for any nested family of subspaces; monotonicity is a generic property of optimal or least-squares approximation, not a sign of barrier removal. The barrier is about the rate of decay: slow, often algebraic, decrease of the n-width as n grows. To show mitigation, the benchmark must demonstrate that the quadratic manifold's error decays substantially faster in n than the best linear subspace error, or that it reaches a target accuracy at a significantly smaller n (with equal or lower total online cost). If the provided evidence is only monotone error curves, the headline claim is not logically established. Additionally, because the supplied full text is corrupted and contains the header of a different paper, the actual benchmark curves and rate comparisons cannot be inspected. This verification gap is not the authors' fault, but it makes the central claim currently unsupported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a unified multi-perspective quadratic manifold reduced-order model (ROM) framework for thermo-mechanically coupled damage-plasticity problems. The abstract claims that by decomposing the solution into multiple physical fields and material states, and by selecting mode counts per field based on material physics, the method achieves a smooth and monotonic decrease in ROM error as the number of modes increases, thereby mitigating the Kolmogorov barrier of linear projection-based ROMs. The full text supplied for review is severely corrupted: it consists largely of unreadable mojibake and begins with the arXiv header of an unrelated astro-ph paper (2508.18221v1). Only the abstract is legible. Consequently, the governing equations, the definition of the quadratic manifold, the mode-selection procedure, and the benchmark results cannot be inspected.","tokens_in":12051,"tokens_out":3951,"duration_ms":45221,"significance":"If the claimed results could be verified, the paper would be a useful contribution to nonlinear model order reduction for damage mechanics, where linear ROMs are known to suffer from slow Kolmogorov n-width decay. The idea of constructing block-separated quadratic manifolds per field/state is plausible. However, the manuscript in its current form provides no verifiable evidence for the central claim: no error curves, no comparison against linear ROMs, no definition of the error metric, and no machine-checked proofs. The conceptual criticism that monotonic error decrease is generic for optimal approximation also applies, so the abstract's success criterion is insufficient to establish barrier mitigation. Therefore the significance is currently potential rather than demonstrated.","major_comments":[{"comment":"The submitted full text is unreadable due to character-encoding corruption and includes the header of an unrelated arXiv paper (2508.18221v1 [astro-ph.HE]). Equations, section numbering, benchmark descriptions, and numerical results cannot be verified. This is a load-bearing issue: the central claim about benchmark performance is entirely unsupported. The authors must provide a clean, complete manuscript before review can proceed.","section":"Full text (entire manuscript)"},{"comment":"The only evidence cited for mitigating the Kolmogorov barrier is 'a smooth and monotonic decrease in error as the number of modes increases.' This is not sufficient: for any nested family of linear subspaces, the Kolmogorov n-width is non-increasing in n, and least-squares projection errors are monotone non-increasing by construction. The barrier is about the rate and saturation of the decay, not monotonicity. To substantiate the claim, the paper must compare the proposed quadratic-manifold error decay against the best linear-subspace benchmark in the same total reduced dimension, and report actual error values and convergence rates for each benchmark. Without such a comparison, the headline claim is a non-sequitur.","section":"Abstract, key-features paragraph"},{"comment":"The 'multi-field and multi-state decomposition strategy grounded in the material's physical response' is stated but no concrete selection rule is given in the abstract, and the full text is unreadable. Per-field mode counts appear to be free parameters selected for each benchmark. If these counts are chosen after seeing the test errors, the monotone error curves may be an artifact of tuning rather than a property of the manifold. A fixed, a-priori selection criterion or a sensitivity study over mode-count choices is required to rule out circularity.","section":"Abstract, mode-selection claim"},{"comment":"The proposed quadratic manifold appears to be built by concatenating per-field, per-state modes. If the resulting manifold is block-diagonal in field components, the cross-coupling between displacement, temperature, and damage is only representable through the quadratic terms of each block. It is not demonstrated that such a block-separated manifold can faithfully represent the coupled solution manifold that drives damage evolution. A formal argument or a targeted numerical experiment (e.g., comparing manifold projection error against a full coupled quadratic manifold) is needed. Currently this point cannot be assessed because of the corrupted text.","section":"Section 3 (manifold construction, based on readable fragments)"}],"minor_comments":[{"comment":"Fix the character encoding and re-submit a clean PDF/TeX source. Remove the extraneous arXiv header from 2508.18221v1.","section":"Full text"},{"comment":"Define the error metric precisely (e.g., relative L2 error in space-time) and report the total reduced dimension, not just per-field mode counts.","section":"Benchmark description"},{"comment":"Include error bars or repeated-run variability, and clarify whether the benchmarks are deterministic. Report wall-clock or online-cost comparisons against the full-order model.","section":"Numerical results"},{"comment":"Add a comparison with a standard linear POD-Galerkin ROM and, if possible, with a standard quadratic manifold without the multi-perspective decomposition, to isolate the benefit of the proposed construction.","section":"Comparisons"}],"recommendation":"major_revision","confidential_remarks":"The corruption of the manuscript is likely a submission or compilation error, not an authorial defect, but as submitted the paper is not reviewable. I recommend requesting a clean copy and, simultaneously, addressing the evidentiary gap: monotone error decrease alone cannot support the Kolmogorov-barrier-mitigation claim. The per-field mode-count selection as a tunable parameter also needs clarification. If the clean version does not add rate comparisons or a fixed mode-selection rule, the central claim may remain unsupported."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The multi-field, multi-state quadratic manifold idea is worth taking seriously for thermo-mechanically coupled damage-plasticity. The physics-guided decomposition into per-field, per-state modes is a genuinely new twist on existing quadratic-manifold ROMs, and it targets a real gap: linear ROMs that stall on damage problems. That alone justifies interest from the ROM and computational-mechanics communities.\n\nBut the abstract's evidence does not establish the headline claim. Monotone error decrease is not the same as mitigating the Kolmogorov barrier. The n-width is non-increasing in dimension for any nested linear family, so smooth monotone decay is a generic property of optimal or least-squares approximation. The barrier is about the rate: you need to show the quadratic manifold reaches a target accuracy at much smaller n than the best linear subspace, with a fair comparison of online cost. The abstract never promises that comparison, so the central claim is currently unsupported.\n\nThe per-field mode-count selection is another soft spot. Choosing mode counts per benchmark, even if physics-guided, is a free parameter. Without a fixed selection rule or validation on a held-out case, the error curves could partly reflect curve fitting rather than a robust property of the manifold. The abstract also gives no error bars or baseline details, so the benchmark evidence is under-specified.\n\nOn top of that, the full text we received is corrupted — it is mostly mojibake and even carries the arXiv header of a different paper (2508.18221, astro-ph.HE). I cannot inspect the equations, figures, or references, so anything beyond the abstract is inaccessible. That is not the authors' scientific fault, but it makes verification impossible.\n\nWho gets value here: readers interested in nonlinear ROMs for coupled damage should watch for a clean version. As it stands, I would not send this to referees; I would ask the authors to fix the text and redo the benchmark analysis with explicit n-width-style rate comparisons and a fixed mode-selection rule. Then it deserves serious refereeing. The idea is plausible, but the evidence is not there yet.","headline":"A plausible new nonlinear ROM idea for coupled damage-plasticity, but the abstract's success criterion is too weak to back the barrier-mitigation claim, and the supplied full text is unreadable.","tokens_in":12591,"tokens_out":3542,"would_cite":false,"duration_ms":47374,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that a multi-perspective quadratic manifold, built field-by-field and state-by-state, breaks the Kolmogorov barrier that stalls linear reduced-order models on coupled damage-plasticity problems.","keywords":["reduced-order models","Kolmogorov barrier","quadratic manifold","damage mechanics","thermo-mechanical coupling","plasticity","nonlinear model reduction","multi-field decomposition"],"falsifier":"A concrete test is to take a coupled damage benchmark with a propagating localization band crossing a thermal gradient and compare the decoupled multi-perspective manifold against a full-state quadratic manifold trained on the full coupled solution. If the decoupled version's error plateaus or oscillates with mode count while the full-state manifold keeps improving, the claim that the multi-perspective splitting defeats the Kolmogorov barrier is refuted. A complementary diagnostic is to check whether the decoupled manifold's tangent space at damage-localization snapshots contains the dominant","tokens_in":11714,"feed_emoji":"⚙️","tokens_out":5752,"duration_ms":64646,"temperature":0.7,"pith_summary":"The paper tries to establish that the Kolmogorov barrier in linear reduced-order models can be overcome for thermo-mechanically coupled damage-plasticity simulations by replacing the single linear subspace with a set of quadratic manifolds, one per physical field and material state. In linear ROMs the solution set's Kolmogorov n-width decays slowly, so the error stops falling once enough modes are added. The authors argue that a multi-perspective quadratic manifold, with mode counts selected from the material's physical response, avoids this plateau and produces a smooth, monotonic error decrease as modes increase. If correct, this makes nonlinear model reduction reliable enough for industrial damage simulations, where linear ROMs currently stall.","feed_headline":"Quadratic manifold lifts the Kolmogorov barrier in damage models","feed_subtitle":"Per-field, per-state modes give smooth error decay where linear reduced-order models stall.","key_machinery":"The central object is the multi-perspective quadratic manifold: a reduced approximation in which each physical field (such as temperature and displacement) and each material state (such as plastic and damage variables) has its own quadratic map from latent coordinates to full-field coordinates, with the maps then assembled into one coupled system. The quadratic terms supply curvature that a linear subspace lacks, and the per-field, per-state splitting prevents one field's features from being averaged out by a global basis. Mode numbers are allocated according to the material's physical response, which is the mechanism intended to keep the manifold aligned with the true solution manifold and","core_discovery":"The central claim is that the Kolmogorov barrier is not intrinsic to coupled damage-plasticity problems but a limitation of linear subspace approximation. The paper's framework maps each physical field and each material state through its own quadratic manifold, then couples them into one reduced system; the mode number per field is chosen on physical grounds rather than by a single global energy criterion. On benchmark problems, this multi-perspective quadratic manifold gives a smooth, monotonic decrease in error with mode count, while the linear projection ROM saturates. The paper presents this as evidence that the multi-perspective splitting captures the cross-field and cross-state interac","pith_inferences":["A sharper diagnostic than the error-vs-modes curve would be to compare the tangent spaces of the true coupled solution manifold with the tangent spaces of the decoupled per-field quadratic manifolds; if the decoupled tangent directions span the coupled ones only in the tested regimes, the smooth decay may not transfer to moving damage zones or crack branching.","The physics-based mode rule could likely be automated by estimating damage-zone width or temperature-boundary-layer thickness a priori, turning a benchmark-informed choice into a transferable criterion.","The same multi-perspective idea may extend to other multiphysics systems whose fields have very different spatial regularity, such as fluid-structure or electro-thermo-mechanical problems, where a single global linear basis mixes scales and loses accuracy."],"forward_implications":["On the benchmark thermo-mechanically coupled damage-plasticity problems, the framework yields ROM error that decreases monotonically as modes are added, instead of reaching the linear ROMs' plateau.","The per-field, per-state decomposition lets an engineer see which field or material state dominates the error budget, guiding where to spend additional modes.","The reduction scheme is agnostic to the exact damage model: the same quadratic-manifold construction can be applied to other damage-involved multiphysics settings with a re-allocation of modes.","Because the online approximation stays quadratic, the computational savings of projection-based ROMs are largely retained while the nonlinear capacity is increased.","The framework turns mode selection into a physics-informed decision rather than a purely algebraic one, which is what the authors argue removes the barrier."],"supporting_citations":[],"fun_headline_variants":["Quadratic manifold ends Kolmogorov barrier in damage models","Per-field quadratic manifolds smooth error decay in damage ROMs","Multi-state quadratic manifold bypasses Kolmogorov barrier","Physics-guided modes give smooth error decay past Kolmogorov barrier"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is that a quadratic manifold assembled from separately reduced fields and states still contains the cross-coupling terms that actually drive damage evolution; if the decoupling discards those interactions, adding modes cannot repair the representation.","fun_headline_variants_meta":{"raw":{"variants":["Quadratic manifold ends Kolmogorov barrier in damage models","Per-field quadratic manifolds smooth error decay in damage ROMs","Multi-state quadratic manifold bypasses Kolmogorov barrier","Physics-guided modes give smooth error decay past Kolmogorov barrier"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000765,"raw_usage":{"total_tokens":3233,"prompt_tokens":754,"completion_tokens":2479,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":498,"completion_tokens_details":{"reasoning_tokens":2411}},"tokens_in":498,"tokens_out":2479,"duration_ms":21864,"temperature":1.0,"reasoning_tokens":2411,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T16:32:08.973277+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete test is to take a coupled damage benchmark with a propagating localization band crossing a thermal gradient and compare the decoupled multi-perspective manifold against a full-state quadratic manifold trained on the full coupled solution. If the decoupled version's error plateaus or oscillates with mode count while the full-state manifold keeps improving, the claim that the multi-perspective splitting defeats the Kolmogorov barrier is refuted. A complementary diagnostic is to check whether the decoupled manifold's tangent space at damage-localization snapshots contains the dominant","supporting_citations":[],"review_version":1}