{"id":"f04a297b-3b8a-4a20-82a9-fa9df9007d85","arxiv_id":"2508.17870","paper_version":3,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A field-aware neural network potential computes polarization, Born charges, and polarizability by exact differentiation of a learned electric enthalpy, reproducing DFT-level dielectric and ferroelectric response across materials.","lead":"This paper builds an upgraded machine-learning model of atoms that can feel an applied electric field, so one model outputs polarization, charges, and dielectric response for many materials at once. It claims accuracy close to expensive quantum calculations for ferroelectric switching in BaTiO3 and the infrared, Raman, and dielectric spectra of quartz, which would speed up materials discovery.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Branch-resolved Berry-phase polarization cannot be represented by a single-valued enthalpy derivative; the abstract's 'same-branch' comparisons may inherit reference branch bookkeeping, and the supplied full text is the wrong paper.","rationale":"The reader's weakest assumption—that a single-valued enthalpy functional can encode branch-resolved Berry-phase polarization—is indeed the most load-bearing technical point. If this representational premise fails, the entire approach of learning electric response via exact differentiation cannot deliver branch-resolved polarization without additional bookkeeping, regardless of how well the model fits the training data. The abstract's wording suggests the authors are aware of the branch issue, but it does not explain how a single-valued F overcomes the multivaluedness of P. This is not a disagreement with the consensus; it is an internal consistency concern about the mathematical mapping from the learned functional to the physical observable. The supplied full text being a different paper compounds the problem: no methods, equations, or benchmark details are available to check whether the authors addressed this. Thus the honest verdict remains UNVERDICTED, with confidence low. No adjustment to the reader's verdict is needed, but the branch-representability question should be resolved before any stronger verdict is issued.","tokens_in":2211,"tokens_out":2502,"duration_ms":32088,"concrete_test":"Obtain the actual MACE-Field manuscript and locate the definition of the polarization loss and the training-data generation protocol. Then directly test representability: take a trained MACE-Field model and compute P = -dF/dE along a path in (R, E) that crosses a polarization quantum in the DFT reference (e.g., a soft-mode displacement in BaTiO3 where P changes by e·a/Ω with a the lattice constant and a the cubic cell parameter). If the model's P remains on the original branch while the DFT labels switch branch, a single smooth F cannot fit both endpoints; check whether the training targets were branch-aligned (e.g., by adding integer polarization quanta or by parallel transport) and whether the reported 'same-branch' comparisons depend on post-hoc branch matching. Also verify that the abstract's conductivity and Raman claims are backed by held-out test sets rather than in-sample fitted","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that a smooth, single-valued electric enthalpy F({R}, E) yields P = -dF/dE, while Berry-phase polarization is defined only modulo a polarization quantum and is branch-dependent. The abstract's careful wording 'same-branch Berry-phase polarisations' and 'branch-resolved polarisation' indicates that the learned model is being compared to a specific branch choice of the DFT reference. If P is obtained by exact differentiation of a scalar F, the model can only produce one continuous branch of polarization. Any reference data that crosses a branch boundary—e.g., a ferroelectric switching path where the physical polarization changes by a polarization quantum—cannot be represented without a discontinuity in F or an unphysical energy jump. The paper must explain how branch alignment is encoded: if branch labels are part of the training targets, the model inherits the reference's branch bookkeeping and the claim of 'learning' polarization is weakened; if no branch labels are used, the model cannot distinguish states that differ by a quantum. This representability concern is distinct from the numerical accuracy claim. Additionally, the supplied full text is arXiv:2508.17864 on type-IV altermagnetism, not the MACE-Field paper, so the architecture details, loss functions, data splits, and benchmark protocols cannot be verified. These two issues jointly leave the central claim uncheckable from the available material.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript, as submitted, consists of an abstract advertising a machine-learned interatomic potential called MACE-Field, claimed to learn an electric enthalpy functional F({R}, E) and to obtain polarization P, Born charges Z*, and dielectric polarizability alpha by exact differentiation, with benchmarks including BaTiO3 hysteresis and alpha-quartz spectra; the full text supplied, however, is a different paper on the symmetry classification of altermagnetism and type-IV magnetism in two dimensions (arXiv:2508.17864). No methods, architecture details, loss functions, data splits, training protocols, or benchmark results for MACE-Field are present in the submitted text. The central claims are therefore not checkable from the submitted material. In addition, the abstract's use of 'same-branch Berry-phase polarisation' raises a representability concern: a single-valued scalar enthalpy functional cannot generally reproduce branch-resolved Berry-phase polarization if the reference data crosses polarization-quantum branch boundaries, unless branch labels are encoded in training or post-processing, which the abstract does not explain.","tokens_in":2559,"tokens_out":1578,"duration_ms":20943,"significance":"If the MACE-Field approach were realized as described, it would constitute a practically important step: a plug-in field coupling for equivariant foundation models yielding dielectric and ferroelectric response from a single learned functional, with DFT/DFPT-level fidelity. Such a capability would be useful for high-throughput screening and finite-field molecular dynamics. However, the submitted manuscript provides none of the supporting evidence. The only full text is an unrelated manuscript on type-IV altermagnetism. The referee cannot evaluate the significance of a paper whose body does not contain the claimed work, and the theoretical concern about branch-resolved polarization from a single-valued derivative further requires a concrete resolution before the claim can be credited.","major_comments":[{"comment":"The submitted full text is not the paper described in the abstract. The body concerns 'Symmetry Classification of Altermagnetism and Emergence of Type-IV Magnetism in Two Dimensions' (apparently arXiv:2508.17864), whereas the abstract introduces MACE-Field, an electric enthalpy functional with exact derivatives, benchmarks on BaTiO3 and alpha-quartz, and a multihead foundation model. None of the methods, datasets, training protocols, or numerical results for MACE-Field appear in the manuscript. The central claims are therefore entirely unsupported by the submitted material. This is not a local revision issue; the manuscript as submitted cannot be evaluated.","section":"Full Text (all sections)"},{"comment":"The abstract states that P is obtained by exact differentiation of a single scalar functional F({R},E), i.e., P = -dF/dE, and reports 'same-branch Berry-phase' polarization comparisons. Berry-phase polarization is multivalued modulo a polarization quantum and is branch-dependent; a differentiable scalar F yields a single continuous branch. If the model cannot represent states differing by a polarization quantum, ferroelectric switching paths or comparisons across branch boundaries are either impossible or inherit reference branch bookkeeping. The manuscript must specify how branch alignment is defined, whether branch labels enter training, and how transitions across quantum boundaries are handled. Without this, the representability claim is not yet established.","section":"Abstract"},{"comment":"The benchmark claims ('comparable to DFPT', reproduction of hysteresis loops and infrared/Raman/dielectric spectra, cross-chemistry polarization trends) are asserted without any error bars, held-out structure lists, or protocol descriptions. Since the submitted text contains no Results or Methods sections, these claims cannot be verified or meaningfully compared with existing DFPT or machine-learning benchmarks.","section":"Abstract (benchmarks)"}],"minor_comments":[{"comment":"The arXiv identifier embedded in the pagination footer (2508.17864) differs from the manuscript's stated identifier (2508.17870), confirming that the supplied full text belongs to another submission. This mismatch should be corrected at the source.","section":"Title/Abstract metadata"},{"comment":"The phrase 'same-branch Berry-phase' is undefined. If the model is trained on branch-resolved labels, that should be stated explicitly; if not, the meaning of 'same-branch' is unclear.","section":"Abstract"}],"recommendation":"reject","confidential_remarks":"To the editor: This is not a case where a flawed but checkable manuscript can be revised. The submitted full text is a completely different paper. Even if the abstract describes a plausible research program, the referee cannot review claims whose supporting material is absent. I recommend desk rejection or a resubmission with the correct full text. Additionally, the branch-resolved polarization issue is substantive and should be addressed in any future submission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this one for the representability question, not for the benchmarks. The new thing is MACE-Field: a plug-in that couples a uniform electric field into latent equivariant features of MACE, then reads P, Z*, alpha out by exact differentiation of a single enthalpy functional. That is a real design step beyond standard MACE and the mace-mp-mh-0 foundation models, and the upgrade path for existing models is useful. The abstract is careful about symmetry and sum rules, which is good.\n\nBut the central claim has a soft spot that the abstract's own wording trips over. Berry-phase polarization is defined modulo a quantum and is branch-dependent. If P = -dF/dE with F single-valued, you get one continuous branch. So 'same-branch Berry-phase polarisations' and 'branch-resolved polarisation' are doing too much work. Either the branch labels are part of the training targets, in which case the model inherits the reference's branch bookkeeping and the 'learning' claim weakens, or they are not, in which case the model cannot represent states that differ by a quantum. That is a representability problem, not a numerical accuracy problem, and the paper needs to explain how branch alignment is encoded. I don't see a way around it from the abstract alone.\n\nI also can't verify the numbers: the supplied full text is arXiv:2508.17864 on altermagnetism, not this paper. So no error bars, no data splits, no benchmark protocols. The headline claims ('comparable to DFPT', cross-chemistry transfer, hysteresis loops) are asserted without statistical detail. The BaTiO3 hysteresis from a single-material model is likely in-sample. That is not a flaw in the paper if the full text provides hold-out protocols, but it is uncheckable here.\n\nThe conceptual issue is the load-bearing one. It deserves a serious referee because the method is plausible and the branch problem is exactly what a referee should probe. If the full text addresses it, this could be a solid contribution; if not, the transferability claims are much weaker than advertised.\n\nMy take: send it to review, but demand a clear statement of how branch-resolved polarization is represented and how the model avoids inheriting reference branch conventions. Also ask for hold-out details and error bars. I'd bring this to a reading group to argue about the Berry-phase point, but I wouldn't cite it until I see the actual methods.\n\nBest.","headline":"A promising field-aware MACE variant with a real representability problem around branch-resolved polarization; unverifiable from the supplied text.","tokens_in":3048,"tokens_out":2641,"would_cite":false,"duration_ms":28263,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A learned electric enthalpy functional, differentiated exactly, yields polarization, Born charges, and polarizability from one MACE-based potential.","keywords":["equivariant neural network potentials","electric enthalpy functional","ferroelectric polarization","Born effective charges","dielectric response","Berry phase","finite-field molecular dynamics","foundation model"],"falsifier":"Run a closed electric-field loop on a ferroelectric whose polarization crosses a polarization quantum, and compare the model's trajectory of P with the DFT Berry-phase reference; if MACE-Field switches branches at a different field or returns a path-dependent polarization inconsistent with the reference, the single-valued enthalpy assumption fails.","tokens_in":2105,"feed_emoji":"⚡","tokens_out":5889,"duration_ms":67430,"temperature":0.7,"pith_summary":"This paper introduces MACE-Field, an atomistic machine-learning potential that treats the electric field as one more input to a learned energy function. From that single electric enthalpy functional, the model obtains polarization, Born effective charges, and dielectric polarizability by exact differentiation, so the response tensors are consistent with the forces and with each other by construction. The paper claims that this simple field coupling, added to an existing equivariant foundation model, transfers across chemistries: it reproduces same-branch Berry-phase and spontaneous polarizations for diverse ferroelectrics, and its fine-tuned foundation version predicts dielectric and ferroelectric trends while retaining force-field accuracy. On individual materials, the model also produces BaTiO3 hysteresis loops and alpha-quartz infrared, Raman, and dielectric spectra from finite-field molecular dynamics at fidelity comparable to density-functional perturbation theory. If the central claim holds, quantitative electric response becomes a by-product of one learned enthalpy rather than a per-material calculation.","feed_headline":"One function reproduces ferroelectric loops and dielectric spectra","feed_subtitle":"MACE-Field derives polarization, Born charges, and polarizability from one field-coupled potential, matching DFPT on key materials.","key_machinery":"The central object is the electric enthalpy functional F({R}, E), learned by an O(3)-equivariant graph neural network (the MACE backbone) whose latent equivariant features couple to a uniform applied electric field. Because the energy readout is a scalar function of positions and field, differentiating it with respect to E yields P = -∂F/∂E; further derivatives give Z* and alpha. The design enforces Maxwell reciprocity (cross-derivatives commute), the acoustic sum rule, and the tensor symmetries of crystals by construction, so the response tensors inherit the symmetries of the energy rather than needing to be imposed.","core_discovery":"On the paper's terms, the central claim is that a physics-informed field coupling can give atomistic foundation models transferable dielectric and ferroelectric response. The specific construction is a field-aware O(3)-equivariant interatomic potential, MACE-Field, which learns a single electric enthalpy functional F({R}, E). The uniform field couples to latent equivariant features inside the MACE backbone, and the scalar energy readout preserves Maxwell reciprocity, the acoustic sum rule, and crystal tensor symmetries by construction. P, Z*, and alpha are then obtained by exact differentiation, so the predicted response tensors are not separate fitted outputs but derivatives of one scalar f","pith_inferences":["The model's branch-resolved polarization predictions must inherit reference branch bookkeeping because physical polarization is defined only modulo a polarization quantum; a revealing stress test would drive the model through a nonpolar phase boundary and check whether the branch assignment stays consistent with the Berry-phase reference.","The same field-coupling construction—uniform field into equivariant latent features, exact differentiation of a scalar enthalpy—should port to other invariant or equivariant architectures, since nothing beyond the backbone is MACE-specific.","Finite-field MD with learned enthalpies could extend to other enthalpy derivatives such as piezoelectric and nonlinear optical coefficients, but those are not demonstrated here.","The foundation model's benchmark success suggests it could predict temperature-dependent dielectric constants or ferroelectric switching paths for new chemistries without retraining, though quantitative DFPT-comparable fidelity outside the benchmark set remains open."],"forward_implications":["A single learned enthalpy functional supplies P, Z*, and alpha as exact derivatives, so the response tensors are internally consistent with forces and with each other; no separate training targets are needed for each tensor.","Existing MACE energy/force foundation models can be upgraded to field-aware behavior through the plug-in coupling, transferring electric response across chemistries without per-material physics.","Finite-field molecular dynamics with MACE-Field reproduces experimental observables—hysteresis loops, infrared, Raman, and dielectric spectra—at accuracy comparable to DFPT for the benchmarked materials (BaTiO3, alpha-quartz).","Cross-chemistry polarization prediction is possible including branch-resolved Berry-phase and spontaneous polarization, meaning a foundation model can act as a surrogate for expensive DFPT polarization calculations.","Single-material fine-tuning remains the route to the most quantitative spectroscopic predictions; the foundation model is a strong prior, not a replacement for material-specific training in high-accuracy regimes."],"supporting_citations":[],"fun_headline_variants":["MACE-Field: one enthalpy functional yields P, Z*, alpha","Differentiate learned enthalpy to get all electric response tensors","MACE-Field: field coupling yields transferable electric response","From one scalar function: ferroelectric loops and dielectric spectra","MACE-Field: one function, all electric responses"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The central assumption is that one smooth, single-valued learned energy function of atomic positions and electric field can carry the branch-resolved information of Berry-phase polarization, even though physical polarization is only defined up to a whole polarization quantum; the branch bookkeeping must come out right.","fun_headline_variants_meta":{"raw":{"variants":["MACE-Field: one enthalpy functional yields P, Z*, alpha","Differentiate learned enthalpy to get all electric response tensors","MACE-Field: field coupling yields transferable electric response","From one scalar function: ferroelectric loops and dielectric spectra","MACE-Field: one function, all electric responses"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000918,"raw_usage":{"total_tokens":3833,"prompt_tokens":861,"completion_tokens":2972,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":605,"completion_tokens_details":{"reasoning_tokens":2886}},"tokens_in":605,"tokens_out":2972,"duration_ms":27598,"temperature":1.0,"reasoning_tokens":2886,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T16:44:52.701139+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a closed electric-field loop on a ferroelectric whose polarization crosses a polarization quantum, and compare the model's trajectory of P with the DFT Berry-phase reference; if MACE-Field switches branches at a different field or returns a path-dependent polarization inconsistent with the reference, the single-valued enthalpy assumption fails.","supporting_citations":[],"review_version":1}