{"id":"641befdb-2693-4967-abdd-f403e383a80a","arxiv_id":"2502.07293","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"By rescaling atomic pair distances with element-pair-specific parameters, the authors make one shared radial function serve all elements, yielding an ultra-small machine learning interatomic potential with accuracy close to large neural network models.","lead":"Researchers built a machine-learning model for atomic forces whose radial functions are rescaled by a universal equation-of-state law, cutting model parameters by orders of magnitude while keeping accuracy close to much larger neural potentials. The model, SUS2-MLIP, is tested on alloys, oxides, semiconductors, half-Heusler thermoelectrics, and battery electrolytes, and is open-sourced.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"UEOS constraint on total energy does not imply universal form for each pair function; Eq. (6) is an ansatz, not a derivation.","rationale":"The paper's strongest claim is the universal radial function that reduces the parameter space and enables physical scalability. The load-bearing point is the derivation in §II.2, where the 'therefore' connecting UEOS to universal pair functions is unsupported. UEOS is a statement about a one-dimensional scalar energy-versus-volume curve, while Eq. (1) contains many degrees of freedom per atom; inferring universality of the pair factors from the total curve is a logical gap unless additional assumptions are stated. The fitted per-pair scaling parameters and the flexible learned radial function make the collapse partly a fitting outcome rather than an independent physical law. This is not a disagreement with community consensus but a correctness risk in the derivation. The empirical benchmarks are extensive and the code is available, so the paper has genuine value as a compact MLIP; however, the central theoretical mechanism should be treated as conditional pending either a rigorous derivation or a direct falsification test. The reader's weakest_assumption identifies the same issue, and the verdict should remain conditional rather than being shifted further.","tokens_in":17381,"tokens_out":4924,"duration_ms":51827,"concrete_test":"Fit a variant of SUS2-MLIP on the alloy benchmark with fully per-pair radial functions R_l,k^{ZI,ZJ}(r) for the same l,k channels, then test whether the fitted per-pair functions collapse onto one shared R̃_l,k after optimizing the per-pair scaling parameters α_{ZI,ZJ} and r0_{ZI,ZJ}. If the post-scaling spread is comparable to the training noise floor, Eq. (6) is not a physics-derived universal law; a complementary analytical check is to re-derive Eq. (6) from Eqs. (1) and (5) without assuming universal φ_IJ, or to construct two distinct families of φ that both yield a UEOS-consistent P_I.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central step is in §II.2, between Eq. (5) and Eq. (6). For an elemental FCC/BCC crystal, Eq. (1) makes P_I(r_nn) a product of pair factors; the paper asserts that because this scalar curve must satisfy UEOS, 'all φ_IJ should also possess the universal representation within the scaling physical space as in equation (5).' That implication is not proved and is not a logical consequence: a constraint on the total normalized cohesive energy as a function of r_nn does not determine the individual factors φ_IJ. Different pair potentials with different radial shapes can be arranged to give the same Rydberg-like P_I(r_nn) over the fitted range, so UEOS alone cannot force each φ_IJ or each many-body term to collapse onto one universal R̃_l(r*). The subsequent parameter reduction from R^{n×l×Z^2} to R^{n×l} in Eq. (6) rests entirely on this unproved universality. The pair-specific α_{ZI,ZJ} and r0_{ZI,ZJ} are themselves fitted parameters, so the apparent collapse is partly self-fulfilling: two affine parameters per pair can absorb much of the chemical variation, and the benchmark accuracy does not independently validate the physical law. If Eq. (6) is only a compact fitting ansatz, the claims of physics-informed global scaling, out-of-domain generalization, and extrapolative scalability are weakened, even though SUS2-MLIP may remain a useful low-parameter MLIP.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes SUS2-MLIP, a low-parameter machine learning interatomic potential built around a generalized-linear cluster expansion. Its central theoretical claim is that the universal equation of state (UEOS) justifies replacing element-pair-specific radial functions with a single element-independent universal radial function R̃_l(r*_IJ) after scaling r*_IJ = α_{ZI,ZJ}(r_IJ/r0_{ZI,ZJ} − 1), reducing the radial parameter space from R^{n×l×Z^2} to R^{n×l}. The model is benchmarked against DPA-2, MACE, Nequip, Allegro, and other models on alloy, cathode, perovskite, and semiconductor datasets, and is applied to phonon/thermal conductivity calculations for half-Heuslers, Cu2Se molecular dynamics, and Li-ion diffusion in sulfide electrolytes. An open-source implementation is provided.","tokens_in":17846,"tokens_out":5522,"duration_ms":53756,"significance":"If the universal scaling of Eq. (6) were established, the parameter reduction and out-of-domain transferability would be a significant practical contribution: the reported parameter counts are genuinely small and the property predictions (phonons, thermal conductivity, ionic diffusivity) are broad and encouraging. The paper also ships code and performs genuine out-of-sample benchmarks. However, the central derivation is not currently established: the UEOS constrains total energy versus volume, not individual pair interaction functions, and the fitted per-pair scaling parameters α and r0 absorb part of the claimed universality. The practical model may still be a useful compact ansatz, but the physics-derived law claim needs substantial revision or supporting evidence.","major_comments":[{"comment":"The inference that UEOS implies every pair interaction function φ_IJ inherits the two-parameter universal scaling is not proved and is not a logical consequence of Eq. (5). For a one-element BCC/FCC crystal, Eq. (1) gives P_I(r_nn) = p_I ∏_j(1+φ_IJ(r_nn)), so the requirement that P_I satisfy the universal Rydberg form constrains only a specific scalar combination of pair functions, mainly the isotropic l=0 channel under cubic symmetry. It does not determine the angular decomposition, the l>0 radial channels, or the heteronuclear pair functions. Eq. (6) is therefore an ansatz, and the central parameter reduction from R^{n×l×Z^2} to R^{n×l} rests on this unproved universality.","section":"II.2, Eq. (5)–Eq. (6)"},{"comment":"The parameter-count comparison omits the fitted scaling parameters α_{ZI,ZJ,η} and r0_{ZI,ZJ,η} from the claimed reduction. These parameters are element-pair- and channel-dependent and grow as O(Z^2), so the statement that the model 'decouples the element space from coordinate space' is overstated. To support the universality claim, the authors should either report the total number of fitted parameters including α and r0 or provide a direct test that pair-specific fitted radial functions collapse onto a common R̃_l(r*) after scaling, rather than inferring the collapse from aggregate benchmark accuracy.","section":"Eq. (6) and Fig. 4"},{"comment":"The semiconductor benchmark is not a controlled comparison. The caption of Fig. 3 and the Methods state that the training and validation sets 'were chosen based on a criterion of max atomic force < 5 eV/Å', while the results for the other MLIP models are taken from ref. 40. If the reference models were trained and evaluated on the full semiconductor dataset, the force RMSE comparison is not apples-to-apples, and the lower-force filtered test set may explain part of the reported accuracy. The baseline models should be retrained on the same filtered set, or the comparison should be reported both on the filtered and full datasets.","section":"III.1 and Methods: semiconductor benchmark"},{"comment":"The claim of 'super-linear expressive capacity' is not established. The paper cites the Hopfield-network memory-capacity results of Krotov and Demircigil et al., but those theorems concern the storage capacity of associative memory models for random patterns, not the approximation power of a tanh-composed Chebyshev radial basis for interatomic potential energy surfaces. The statement that 'tanh enables R̃ to possess a super-polynomial expansion form' does not by itself imply super-linear approximation capacity. The authors should either provide a concrete approximation-theoretic statement for the class of radial functions relevant to PES fitting or soften the super-linearity claim to a phenomenological one.","section":"II.3, Eq. (8)"}],"minor_comments":[{"comment":"The definition of the scaled coordinate is inconsistent between the main text and the supplementary information: Eq. (6) defines r*_IJ = α_{ZI,ZJ}(r_IJ/r0_{ZI,ZJ} − 1), while Eq. (S2) in Supplementary S2 defines r*_{IJ,η} = α_{ZI,ZJ,η}(r_IJ − r0_{ZI,ZJ,η}). The authors should use one definition throughout and clarify the units of r0.","section":"Eq. (6) vs. Supplementary S2"},{"comment":"There are numerical inconsistencies in the reported parameter counts: the text lists 2,130 parameters for the cathode model and 6,092 for the perovskite model, whereas the Methods lists 2,066 and 6,028, respectively. These should be reconciled.","section":"III.1, Methods"},{"comment":"The labels 'DTF' in Fig. 5(b)–(c) and the associated text should read 'DFT', and the abstract contains typographical errors such as 'outcomes' (for 'overcomes' or 'avoids') and 'enbeding' (for 'embedding').","section":"Fig. 5 and general text"},{"comment":"The quantities τ_I and τ_j are introduced as element-specific parameters but their contribution to the total parameter count is not discussed. Since the paper is explicitly about ultra-small parameterization, the authors should state whether these parameters are fitted and include them in the parameter counts.","section":"II.1, Eq. (4)"},{"comment":"The benchmark figure reports single error values without repeated-seed statistics or error bars. Reporting the mean and standard deviation over multiple training runs would strengthen the comparison, especially given the very small model sizes involved.","section":"III.1, Fig. 3"}],"recommendation":"major_revision","confidential_remarks":"The practical model appears promising and the empirical coverage is broad, but the theory section currently overclaims a derivation from UEOS. The core issue is fixable within the manuscript's scope: reframe Eq. (6) as an empirical universal-scaling ansatz and provide a direct test of the collapse (for example, by fitting pair-specific radial functions and comparing them to the shared R̃_l after scaling), or provide a rigorous derivation from the many-body cluster expression. The semiconductor benchmark comparison also needs to be made controlled. If these points are addressed, the paper could become a solid contribution to compact and transferable MLIPs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know: this is a serious empirical paper with a shaky theoretical center. The authors build a generalized linear MLIP (ACE/MTP-style) and claim the universal equation of state lets them replace element-pair radial functions with one universal radial function after per-pair scaling. That step, between Eq. (5) and Eq. (6), is not proven—UEOS constrains total energy vs volume, not each pair interaction. The stress-test note is right. But the paper still does a lot well.\n\nFirst, the empirical work is substantial. They train on alloy, cathode, perovskite, semiconductor, half-Heusler, Cu2Se, and sulfide electrolyte datasets, and report accuracy comparable to DPA-2, MACE, Nequip, etc., with 2–3 orders of magnitude fewer parameters (17k–29k vs. millions). They also demonstrate physical scalability: phonon dispersions and lattice thermal conductivity for 130 half-Heusler compounds, MD-based thermal transport in beta-Cu2Se, and Li diffusion in LGPS-type electrolytes, with comparisons to DFT and experiment. The open-source implementation is a plus. These are genuine out-of-sample tests, not just fits.\n\nThe soft spots are real but proportionate. The central theoretical claim is overreaching: the UEOS does not force every pair function onto one universal curve. The per-pair scaling constants alpha and r0 are themselves fitted parameters, so part of the apparent collapse is an affine transformation absorbing chemical variation. That makes Eq. (6) a compact fitting ansatz rather than a physics-derived law. The benchmark on semiconductors also filters frames with max force < 5 eV/Å, which could bias the comparison toward smaller forces; that’s a minor concern given the other benchmarks, but worth flagging. The super-linearity argument leans on Hopfield/KAN analogies and is mostly hand-waving, though it’s not load-bearing for the empirical claims.\n\nWho is this for? People working on universal or low-parameter MLIPs, and anyone who cares about whether physics-inspired constraints can reduce model size. The paper deserves a serious referee. I’d recommend major revision: reframe the universality claim as an inspired ansatz rather than a derivation, add a test that checks whether the universal radial function actually transfers to element pairs unseen in training, and be clearer about what the fitted alpha/r0 absorb. That would make the contribution both honest and stronger.","headline":"A compact MLIP that works better than its own derivation justifies; the universal-scaling step is an ansatz, not a theorem, but the empirical case is strong enough to referee.","tokens_in":18234,"tokens_out":1483,"would_cite":true,"duration_ms":15483,"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":"An element-independent universal radial function can replace element-pair-specific fits in machine-learned interatomic potentials, collapsing parameter counts by two to three orders of magnitude while retaining comparable accuracy across…","keywords":["machine learning interatomic potentials","universal equation of state","universal scaling","potential energy surface","ultra-small parameterization","phonon transport","ionic diffusion","super-linear expressivity"],"falsifier":"Train the model on one-element crystals only, read out the learned scaled radial functions $\\tilde R_l(r^*)$ for different element pairs, and check whether they overlap on a common curve; if they do not, the universal-radial claim is refuted. A complementary test is to train only near equilibrium and then evaluate on strongly compressed or expanded configurations: if energy and force errors grow as fast as an unconstrained linear model, then the equation-of-state constraint is not doing the extrapolation work claimed.","tokens_in":17199,"feed_emoji":"⚛️","tokens_out":9034,"duration_ms":73628,"temperature":0.7,"pith_summary":"This paper argues that a machine-learning interatomic potential can keep full many-body expressive power while replacing the usual element-pair-specific radial functions with a single element-independent universal radial function. The replacement succeeds because each pair distance $r_{IJ}$ is first rescaled by two element-pair parameters into a scaled coordinate $r^*_{IJ}$, and the paper derives this scaling from the universal equation of state. The consequence, if the derivation holds, is that the parameter count stops growing with the square of the number of elements and shrinks by two to three orders of magnitude relative to large neural-network potentials, while energy, force, and stress accuracy remain comparable. The paper demonstrates this compact model on alloys, cathodes, perovskites, semiconductors, half-Heusler thermoelectrics, and superionic conductors, including predictions of phonons, lattice thermal conductivity, Cu-ion diffusion, and Li-ion migration.","feed_headline":"Universal scaling shrinks atomistic ML model parameters by ~1000x","feed_subtitle":"Element-pair radial functions collapse onto one universal curve; phonons, conductivity, and Li diffusion match DFT.","key_machinery":"The load-bearing object is the universal radial function $\\tilde R_l(r^*_{IJ})$, defined by $r^*_{IJ} = \\alpha_{Z_I Z_J}(r_{IJ}/r_0^{Z_I Z_J} - 1)$, where $\\alpha$ and $r_0$ are the two element-pair scaling parameters inherited from the universal equation of state. The cluster property field $P_I(\\boldsymbol r_I;\\{\\boldsymbol r_j\\}) = p_I \\prod_{j\\neq I}(\\varphi_{Ij}+1)$ supplies the many-body expansion, with all $n$-body terms generated from the same pair function $\\varphi_I = \\sum_j \\varphi_{Ij}$. Moment-tensor angular descriptors $r^{\\otimes l}$ provide rotational features, while the nonlinear map $x = \\tanh(r^*)$ sends the scaled distance into the Chebyshev domain $(-1,1)$ and gives the basis super-linear expressive capacity. Together these pieces decouple the element space from the coordinate space, so the learning parameters are a small fixed set of radial coefficients and linear expansion coefficients rather than element-pair-indexed lookup tables.","core_discovery":"The central claim is that the element-dependent radial function $R_l(r_{IJ}, Z_I, Z_J)$ in the cluster expansion of the potential energy surface can be replaced by $R_l \\to \\tilde R_l(r^*_{IJ})$ with $r^*_{IJ} = \\alpha_{Z_I Z_J}(r_{IJ}/r_0^{Z_I Z_J} - 1)$, so the radial basis is no longer indexed by element pairs. The justification is that the cluster property $P_I(r_{\\mathrm{nn}})$ of an isotropic one-element crystal is an equation of state, so it must obey the universal equation of state $E^* = -(r^*+1)e^{-r^*}$, and the paper asserts this forces every pair interaction function $\\varphi_{IJ}$ to live in the same scaled space. With this step, the radial expansion coefficient space shrinks from $\\mathbb{R}^{n\\times l\\times Z^2}$ to $\\mathbb{R}^{n\\times l}$, which is the ultra-small parameterization. The paper then feeds the scaled coordinate through a $\\tanh$ map into Chebyshev polynomials, a nonlinearity-embedded radial function claimed to give super-linear expressive capacity, and shows the resulting model matches larger neural-network potentials in accuracy while using far fewer parameters.","pith_inferences":["Editorial inference: if the universal radial claim is exact, then pair-dimer binding curves of diverse elements should collapse onto one master curve after the two-parameter rescaling; checking this directly would separate a physical law from a flexible fitting ansatz.","Editorial inference: the same scaling argument could be imported into graph-neural-network potentials as a physics prior that initializes or constrains their radial embeddings, reducing their parameter count without changing their update rules.","Editorial inference: a practical testable extension is to train the model on a handful of single-element crystals and then predict multi-element alloys with no additional training; success would demonstrate that the element-pair scaling parameters alone carry the chemical transferability.","Editorial inference: because the model is analytic and linear in its coefficients, its scaled radial functions can be read out directly, offering a way to compare learned interatomic interactions against measured pair potentials across the periodic table."],"forward_implications":["Adding a new element to a multi-element model does not add new radial basis parameters; only the new pair-specific scaling constants $\\alpha$ and $r_0$ need to be supplied.","A model trained mostly near equilibrium can extrapolate to strongly compressed or expanded states because the universal equation of state imposes the correct global volume dependence.","Phonon frequencies, third-order force constants, and lattice thermal conductivity can be obtained from the compact model's potential energy surface with errors comparable to DFT-level references.","Transfer learning works with very few structures: adding 45 configurations of a previously absent element pair corrected the phonon dispersion of a solid solution.","The light parameterization permits molecular dynamics on systems exceeding one million atoms on conventional CPU resources, where a larger neural-network baseline ran out of GPU memory below one hundred thousand atoms."],"supporting_citations":[{"why":"Supplies the universal equation of state whose two-parameter scaling fixes the definition of $r^*$.","marker":"[21]"},{"why":"Extends the universal binding-energy relation to all atomic systems, justifying the global scaling premise.","marker":"[31]"},{"why":"Introduces the adaptive-memory associative-memory idea that motivates embedding nonlinear transformations into the radial basis.","marker":"[33]"},{"why":"Shows that learnable activation functions of sum-of-univariate form can approximate multivariate functions, used to argue the nonlinear radial map expands expressive capacity.","marker":"[35]"},{"why":"Provides the software framework used for moment-tensor-style regression and basis handling in which the model is implemented.","marker":"[32]"},{"why":"Supplies benchmark energy and force errors and parameter counts of large neural-network baselines used for comparison.","marker":"[40]"},{"why":"Supplies the periodic-table dataset and reported accuracy and parameter counts used for the parameter-efficiency comparison.","marker":"[25]"},{"why":"Supplies the half-Heusler database and DFT phonon references used to validate phonon and thermal-conductivity predictions.","marker":"[42]"},{"why":"Supplies the sulfide-electrolyte training set and reference AIMD and baseline diffusivities used in the Li-ion migration tests.","marker":"[48]"},{"why":"Provides the experimental Cu2Se thermal-conductivity data used to validate liquid-like transport simulations.","marker":"[44]"}],"fun_headline_variants":["Universal scaling shrinks MLIP parameters 1000x","Physics-informed scaling gives ML potentials ultra-small size","Global curve collapse crunches ML interatomic model","Super-linear MLIP: one curve replaces element pairs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the universal equation of state for the total cluster energy transfers to each individual pair interaction $\\varphi_{IJ}$, so all element-pair radial functions collapse onto one universal curve after rescaling by $\\alpha$ and $r_0$; a constraint on total energy versus volume does not by itself force every pair basis function to follow that same curve.","fun_headline_variants_meta":{"raw":{"variants":["Universal scaling shrinks MLIP parameters 1000x","Physics-informed scaling gives ML potentials ultra-small size","Global curve collapse crunches ML interatomic model","Super-linear MLIP: one curve replaces element pairs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000185,"raw_usage":{"total_tokens":1377,"prompt_tokens":1053,"completion_tokens":324,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":669,"completion_tokens_details":{"reasoning_tokens":262}},"tokens_in":669,"tokens_out":324,"duration_ms":3642,"temperature":1.0,"reasoning_tokens":262,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T13:11:18.205872+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Train the model on one-element crystals only, read out the learned scaled radial functions $\\tilde R_l(r^*)$ for different element pairs, and check whether they overlap on a common curve; if they do not, the universal-radial claim is refuted. A complementary test is to train only near equilibrium and then evaluate on strongly compressed or expanded configurations: if energy and force errors grow as fast as an unconstrained linear model, then the equation-of-state constraint is not doing the extrapolation work claimed.","supporting_citations":[{"cited_title":"H., Smith, J","cited_arxiv_id":null,"evidence_quote":"Supplies the universal equation of state whose two-parameter scaling fixes the definition of $r^*$."},{"cited_title":"& Smith, J","cited_arxiv_id":null,"evidence_quote":"Extends the universal binding-energy relation to all atomic systems, justifying the global scaling premise."},{"cited_title":"A new frontier for Hopfield networks","cited_arxiv_id":null,"evidence_quote":"Introduces the adaptive-memory associative-memory idea that motivates embedding nonlinear transformations into the radial basis."},{"cited_title":"S., Gubaev, K., Podryabinkin, E","cited_arxiv_id":null,"evidence_quote":"Provides the software framework used for moment-tensor-style regression and basis handling in which the model is implemented."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the half-Heusler database and DFT phonon references used to validate phonon and thermal-conductivity predictions."},{"cited_title":"A Pre-trained Deep Potential Model for Sulfide Solid Electrolytes with Broad Coverage and High Accuracy","cited_arxiv_id":"2406.18263","evidence_quote":"Supplies the sulfide-electrolyte training set and reference AIMD and baseline diffusivities used in the Li-ion migration tests."},{"cited_title":"Copper ion liquid-like thermoelectrics","cited_arxiv_id":null,"evidence_quote":"Provides the experimental Cu2Se thermal-conductivity data used to validate liquid-like transport simulations."}],"review_version":1}