Pith. sign in

REVIEW 3 major objections 5 minor 70 references

SuperSalt: Equivariant Neural Network Force Fields for Multicomponent Molten Salts System

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read SuperSalt: one neural-network potential covers all 11-cation chloride melts at near-DFT accuracy.

desk verdict Strong, useful MLIP paper; transferability claim is solid for tested compositions but rests on a shared MACE-MP0 generator that deserves an independent AIMD-configuration check. read the letter →

arxiv 2412.19353 v1 pith:542OHEWY submitted 2024-12-26 cond-mat.mtrl-sci

Chen Shen , Siamak Attarian , Yixuan Zhang , Hongbin Zhang , Mark Asta , Izabela Szlufarska , Dane Morgan This is my paper · ORCID
classification cond-mat.mtrl-sci
keywords moltensaltsmachinelearninginteratomicpotentialequivariantneuralnetworkMACEtransferabilityactiveBayesianoptimizationchloridemelts
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that a single machine-learned interatomic potential can cover the entire liquid-phase composition space of an 11-cation chloride salt family, not just one or two specific melts. By training a MACE equivariant neural network on configurations drawn only from unary, binary, and 11-component systems, the authors claim near-DFT accuracy for density, bulk modulus, radial distribution functions, heat capacity, and thermal expansion, with force errors roughly an order of magnitude below the universal MACE-MP0 potential on the same tests. If correct, this would replace the current practice of fitting a new potential for every salt composition with one reusable model, and it would make large-scale screening of salt compositions practical. The paper also demonstrates that coupling this potential with Bayesian optimization can locate compositions with target densities after only a handful of molecular dynamics runs.

What carries the argument

The central object is the MACE (multilayer atomic cluster expansion) architecture, an equivariant message-passing neural network that builds many-body atomic descriptors from tensor products of spherical basis functions; with two layers, lmax=3, and 4-body messages it maps local atomic environments to energies and forces. Around it sits a data-generation workflow: MACE-MP0-driven melt-quench MD generates raw configurations for 1-, 2-, and 11-component salts; HDBSCAN active learning selects a few hundred diverse structures per subsystem; and PBE-D3 DFT labels roughly 70,000 structures (about 7 million atoms). The paper's efficiency claim rests on the assumption that salt physics is dominated by pairwise electrostatics, so 1- and 2-component data capture the essential interactions and a small amount of 11-component data teaches the potential to handle many-element environments.

What would settle it

Take a ZrCl4-rich multicomponent melt (or add a 12th cation not in the training set), generate configurations from long molecular dynamics runs, and compare SuperSalt forces and predicted densities and heat capacities against fresh AIMD (PBE-D3) calculations; if force RMSEs on Zr environments exceed the roughly 25 meV/Å range seen for random multicomponent tests, the claimed transferability is limited. A simpler check is to compute SuperSalt's force error on configurations from a melt-quench trajectory of pure ZrCl4 at 1200 K, since Zr4+ already shows the largest per-element force errors in the paper's parity plots.

Watch

Extended reading notes

Core claim

SuperSalt, a MACE potential fitted to DFT (PBE-D3) data for 11 chloride salts (LiCl, NaCl, KCl, RbCl, CsCl, MgCl2, CaCl2, SrCl2, BaCl2, ZnCl2, ZrCl4), achieves near-DFT accuracy for energies and forces across the $2^{11}$ composition space, including ternary and random multicomponent mixtures never seen in training. Energy RMSEs are 0.5 meV/atom on training and validation sets, 0.6 meV/atom on 3300 ternary configurations, and 1.3 meV/atom on 800 random multicomponent configurations; force RMSEs range from 13.7 to 24.4 meV/Å. Predicted densities deviate from AIMD by less than 2% and from experiment by about 5%, heat capacities by up to 4.8%, and thermal expansion coefficients by about 6.7%. Relative to the general-purpose MACE-MP0 foundation model, SuperSalt cuts energy errors by roughly 40–90 times and force errors by roughly 7–10 times on molten salt tests.

Load-bearing premise

The raw configurations fed to the DFT labels are generated by the MACE-MP0 universal potential, which the paper itself says may be inaccurate for molten salts; if that generator never visits some chemically important liquid structure, especially around high-valence Zr4+ and Zn2+ ions, the training labels cannot teach the potential about it.

Editorial extensions

If this is right

  • A single SuperSalt potential replaces dozens of system-specific MLIP fittings for chloride melts, since it reliably predicts all 165 ternary and random multicomponent compositions from training on only 1-, 2-, and 11-component systems.
  • Compositions never present in training, including all ternary mixtures, are predicted at near-DFT accuracy with force RMSEs below about 25 meV/Å.
  • Bayesian optimization on top of SuperSalt-MD finds target-density compositions in as few as six iterations, which the paper argues is impractical with empirical or ab initio methods at this scale.
  • Extending the approach to a 12th element requires only adding one unary, its 12 binary systems, and active-learned 12-component configurations, per the authors' stated plan.
  • The potential enables nanosecond-scale molecular dynamics of multicomponent melts with near-DFT force accuracy, opening the way to screening properties like viscosity that are too expensive for direct AIMD.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same unary-plus-binary-plus-high-order training strategy may transfer to other chemically similar liquid families, such as fluoride or bromide melts, if pairwise electrostatics dominate there as well.
  • A sharper test of transferability would be to hold out entire binary subsystems from training and check whether SuperSalt still predicts them; the reported Test1 and Test2 sets do not fully isolate this.
  • The 5% density deviation from experiment likely reflects systematic DFT-D3 error rather than potential error, so coupling SuperSalt with empirical density corrections could improve absolute property predictions.
  • The HDBSCAN active-learning pipeline could be reused to build compositional foundation potentials for other liquid electrolytes or oxide melts where enumerating all compositions is infeasible.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript presents SuperSalt, a MACE-based equivariant neural network interatomic potential for 11-cation chloride melts (Li, Na, K, Rb, Cs, Mg, Ca, Sr, Ba, Zn, Zr). Training configurations are generated by MACE-MP0-driven MD for unary, binary (at three compositions per pair), and 11-component systems, then selected via HDBSCAN active learning and labeled with PBE-D3 DFT; the final database contains roughly 70,000 structures. The authors report low energy/force RMSEs on training, validation, and two held-out sets (a 3300-structure ternary set and an 800-structure random multicomponent set), and they compare density, RDFs, heat capacity, and thermal expansion against AIMD and some experimental data. A Bayesian optimization workflow is demonstrated for targeting compositions with desired densities. The central claim is that a single potential trained on 1-, 2-, and 11-component data transfers to all intermediate compositions across the 11-cation space with near-DFT accuracy.

Significance. If the transferability claim holds, this would be a practically useful advance: a single MLIP covering a broad molten-salt family could replace many system-specific potentials and accelerate composition screening. The paper's strengths include a large and documented training set, a reproducible active-learning workflow, explicit property comparisons against AIMD, and a concrete application to Bayesian optimization. The reported energy and force errors are low, and the property comparisons show reasonable agreement. However, the central transferability claim rests on an assumption that the MACE-MP0-generated configurations adequately sample the liquid configurational space of all 11 cations, especially the high-valence Zr and Zn environments that already show the largest force errors. This assumption is not demonstrated, and because the held-out test sets are generated with the same MACE-MP0 workflow, the reported RMSEs may not be representative of the full compositional space. The paper is well within the scope of the journal and the community will likely find the dataset and workflow useful, but the strength of the claims currently exceeds the evidence.

major comments (3)
  1. [Methods §MD workflows; Results §Training and testing results] The raw training and test configurations are all generated by MACE-MP0-driven MD, a model the paper itself states 'may not be accurate enough for predicting thermophysical properties of molten systems.' The Test1 and Test2 sets are generated the same way ('similar to what was done for generating training data'), so the held-out errors share the generator's configurational biases. If MACE-MP0 under-samples Zr-rich or Zn-rich liquid structures—the elements with the largest force errors in Figure 2 (third column)—active learning can only select from that biased distribution, and DFT labels cannot recover the unsampled regions. To support the transferability claim, please provide evidence that the MACE-MP0 sampling covers the relevant liquid configurational space, for example by comparing MACE-MP0-generated coordination statistics or RDFs with AIMD for ZrCl4- and ZnCl2-rich compositions, or by adding DFT-labeled configurations from independent sampling (AIMD, classical force fields, or other generators) to both training and test sets.
  2. [Results §Density, Heat capacity, and Thermal expansion; Figure 4] The property comparisons against AIMD use the same PBE-D3 functional and similar ~100-atom cells as the training labels, so they demonstrate consistency with the training electronic-structure method rather than independent physical accuracy. The experimental comparison in Figure 4b covers only six systems and shows an average density deviation near 5%. More importantly, the AIMD validation set includes only one Zr-containing composition (the 11-component Li2Na4Mg3K5Ca9ZnRb3Sr2Zr6Cs2Ba2Cl74) and one Zn-containing ternary (0.58NaCl-0.12CaCl2-0.3ZnCl2), both at low concentrations of the high-valence cations. The claim of transferability to untested compositions, especially 4- to 10-component and Zr/Zn-rich systems, is therefore not fully supported. Please validate SuperSalt on a broader set of compositions that systematically vary the Zr and Zn fractions, or temper the transferability claim to the compositions actually tested.
  3. [Results §Comprehensive workflow; Results §Training and testing results] The training set contains only 1-, 2-, and 11-component systems, and the ternary test set uses a single composition A0.33B0.33C0.34 per ternary system. The statement that a potential trained on 11 elements 'inherently describes all 2048 suballoys' and 'exhibits excellent transferability' across all intermediate compositions is an extrapolation from 1-, 2-, and 11-component data. The paper does not report how many distinct compositions appear in Test2, nor how the 800 random configurations are distributed across 4-, 5-, ..., 10-component systems. Please provide error statistics broken down by number of components and by composition region (e.g., Zr-rich, Zn-rich, alkali-rich) for both Test1 and Test2, so that the transferability claim is quantified rather than asserted.
minor comments (5)
  1. [Results §Training and testing results] There is a typo 'Tes 2' in the sentence describing the Test2 dataset; it should read 'Test 2.'
  2. [Figure 5 caption] The caption contains 'T arget one' and 'T arget two'; these should be 'Target one' and 'Target two.'
  3. [Methods §DFT computations] The names 'V ASP 6.4.2' and 'PA W-PBE' have unintended spaces; they should read 'VASP 6.4.2' and 'PAW-PBE.'
  4. [Results §Comprehensive workflow] The statement 'the entire melt-quench region was mapped to the SuperSalt model with only 2% of configuration space (initial structures)' is unclear and appears inconsistent with the later statement that ~70,000 configurations were selected from more than 20,000,000 raw configurations (about 0.35%). Please clarify what the 2% refers to.
  5. [Methods §MD workflows; Figure S1] The comparison of SuperSalt to MACE-MP0 in Figure S1 should specify exactly how the DFT-D3 correction was applied to MACE-MP0 predictions, since MACE-MP0 was trained on PBE data without D3. Without this detail, readers cannot assess whether the comparison is a fair test of the universal potential's accuracy for molten salts.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular reduction: the thermophysical property predictions are emergent from a DFT-fitted MLIP, though the shared MACE-MP0 generator limits the independence of the transferability test.

full rationale

The claimed derivation is a standard MLIP pipeline: MACE-MP0 MD generates raw configurations; DFT (PBE-D3) supplies energies, forces, and stresses; a MACE model is fit to those labels; density, bulk modulus, RDFs, Cp, and thermal expansion are then obtained from MD with the fitted potential. None of these properties is a training target, and the held-out Test1/Test2 sets contain compositions (ternaries and random multicomponent mixtures) not present in the 1/2/11-component training set. The comparison against AIMD and experimental densities provides an external check. The only caveats are methodological rather than circular: (i) the Methods state that MACE-MP0 'may not be accurate enough for predicting thermophysical properties of molten systems,' yet it generates the raw pool, and the test sets are produced 'similar to what was done for generating training data,' so train and test share the generator's configurational bias; (ii) AIMD validation uses the same PBE-D3 functional that labeled the training data. These facts weaken the strength of the transferability claim but do not make any equation or fitted parameter equivalent to the target by construction. Self-citations (refs 12, 13, 53) are prior external studies, including an independent NaCl-MgCl2 case study used to motivate the binary-composition sampling; they are not invoked as a uniqueness theorem or as the sole justification for the central result.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim is an empirical ML capability claim rather than a derivation. It rests on hand-chosen training, architecture, and active-learning choices (listed above), on the assumption that DFT-PBE-D3 is the right ground truth, and on the assumption that a 1/2/11-component training set spans the full composition space. No new physical entities are introduced. The main non-negotiable input is the quality and coverage of the DFT-labeled configurations; the paper provides indirect evidence (Test1 and Test2) but not the data itself.

free parameters (4)
  • MACE neural network weights and architectural hyperparameters = Not enumerated; model trained to minimize energy and force RMSE
    All network weights are fitted to DFT energies and forces. Architectural choices (2 layers, lmax=3, 4-body messages, 64 channels, 8 A cutoff, 128x0e+128x1o irreps, learning rate 0.001) are hand-chosen and not ablated, so accuracy depends on them.
  • Training data composition mix = Approximately 1/3 unary, 1/3 binary, 1/3 11-cation; binary x=0.25, 0.50, 0.75; 40 eleven-cation compositions
    The decision to train only on unary, binary, and 11-cation systems, and the specific binary compositions, is a modeling choice used to achieve transferability. It is justified by prior author work (ref 53) and by held-out tests, not derived.
  • Active learning selection ratio = 70,000 structures selected from over 20,000,000 raw configurations (about 2% of configuration space)
    HDBSCAN-based clustering and sequential sampling determine which configurations receive DFT labels. The stopping threshold for desired accuracy is not quantified, so the 2% coverage is an ad hoc choice.
  • BO acquisition and sampling hyperparameters = Exploration coefficient 0.1; 10 initial compositions; 10 candidates per iteration; 10% coarse grid; top 5,000 PSO seeds
    These hyperparameters control the Bayesian optimization demonstration but not the SuperSalt potential itself. They are chosen by the authors and not optimized.
assumptions (5)
  • domain assumption PBE-D3 DFT with 700 eV cutoff and 1x1x1 k-point mesh is an accurate reference for molten chloride salts.
    Used for all training labels and AIMD comparisons (Methods, DFT computations). If this reference is inconsistent, both SuperSalt and AIMD share the error.
  • domain assumption Salt physics is dominated by pairwise electrostatics, so 1- and 2-component fits capture much of the essential physics.
    Stated in Results; justifies the 1/2/11 training design. Higher-order interactions are assumed to be covered by 11-component data.
  • domain assumption MACE-MP0-driven MD generates representative liquid configurational space despite being fitted to crystals.
    Methods MD workflows: raw configurations are generated by MACE-MP0. If it under-samples high-valence or multicomponent environments, the DFT-labeled training set is biased.
  • ad hoc to paper Configurations selected from 1-, 2-, and 11-component systems generalize to all intermediate compositions (3 to 10 components).
    This is the transferability premise of the workflow, tested on Test1 and Test2 but not proven for all 2048 combinations.
  • domain assumption HDBSCAN active learning selects uncorrelated, representative configurations with only 2% of configuration space.
    The paper assumes the clustering algorithm covers the melt-quench region; no detailed validation of cluster completeness is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SuperSalt: Equivariant Neural Network Force Fields for Multicomponent Molten Salts System." pith.science (2026). https://pith.science/paper/542OHEWY

@misc{pith2026241219353,
  author       = {Pith},
  title        = {Pith review of: SuperSalt: Equivariant Neural Network Force Fields for Multicomponent Molten Salts System},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/542OHEWY}},
  note         = {Machine review of arXiv:2412.19353}
}
read the original abstract

Molten salts are crucial for clean energy applications, yet exploring their thermophysical properties across diverse chemical space remains challenging. We present the development of a machine learning interatomic potential (MLIP) called SuperSalt, which targets 11-cation chloride melts and captures the essential physics of molten salts with near-DFT accuracy. Using an efficient workflow that integrates systems of one, two, and 11 components, the SuperSalt potential can accurately predict thermophysical properties such as density, bulk modulus, thermal expansion, and heat capacity. Our model is validated across a broad chemical space, demonstrating excellent transferability. We further illustrate how Bayesian optimization combined with SuperSalt can accelerate the discovery of optimal salt compositions with desired properties. This work provides a foundation for future studies that allows easy extensions to more complex systems, such as those containing additional elements. SuperSalt represents a shift towards a more universal, efficient, and accurate modeling of molten salts for advanced energy applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

70 extracted references · 63 canonical work pages

  1. [1]

    Fumi, F., Tosi, M.: Ionic sizes and born repulsive parameters in the NaCl-type alkali halides-I: The Huggins-Mayer and Pauling forms. J. Phys. Chem. Solids. 25(1), 31–43 (1964)

    work page 1964

  2. [2]

    Sangster, M., Dixon, M.: Interionic potentials in alkali halides and their use in simulations of the molten salts. Adv. Phys. 25(3), 247–342 (1976)

    work page 1976

  3. [3]

    Faraday Discuss

    Aguado, A., Bernasconi, L., Jahn, S., Madden, P.A.: Multipoles and interaction potentials in ionic materials from planewave-DFT calculations. Faraday Discuss. 124, 171–184 (2003)

    work page 2003

  4. [4]

    Salanne, M., Rotenberg, B., Jahn, S., Vuilleumier, R., Simon, C., Madden, P.A.: Including many-body effects in models for ionic liquids. Theor. Chem. Acc. 131, 1–16 (2012)

    work page 2012

  5. [5]

    Hohenberg, P., Kohn, W.: Inhomogeneous electron gas. Phys. Rev. 136(3B), 864 (1964)

    work page 1964

  6. [6]

    Kohn, W., Sham, L.J.: Self-consistent equations including exchange and correla- tion effects. Phys. Rev. 140(4A), 1133 (1965)

    work page 1965

  7. [7]

    Bengtson, A., Nam, H.O., Saha, S., Sakidja, R., Morgan, D.: First-principles molecular dynamics modeling of the LiCl–KCl molten salt system. Comput. Mater. Sci. 83, 362–370 (2014)

    work page 2014

  8. [8]

    Nam, H.O., Morgan, D.: Redox condition in molten salts and solute behavior: A first-principles molecular dynamics study. J. Nucl. Mater. 465, 224–235 (2015)

    work page 2015

Show all 70 references
  1. [9]

    Li, X., Song, J., Shi, S., Yan, L., Zhang, Z., Jiang, T., Peng, S.: Dynamic fluc- tuation of U 3+ coordination structure in the molten LiCl-KCl eutectic via first principles molecular dynamics simulations. J. Phys. Chem. A 121(3), 571–578 (2017)

    2017

  2. [10]

    Behler, J., Parrinello, M.: Generalized neural-network representation of high- dimensional potential-energy surfaces. Phys. Rev. Lett. 98(14), 146401 (2007)

    2007

  3. [11]

    ACS Appl

    Lam, S.T., Li, Q.-J., Ballinger, R., Forsberg, C., Li, J.: Modeling LiF and FLiBe molten salts with robust neural network interatomic potential. ACS Appl. Mater. Interfaces 13(21), 24582–24592 (2021)

    2021

  4. [12]

    Attarian, S., Morgan, D., Szlufarska, I.: Studies of Ni-Cr complexation in FLiBe molten salt using machine learning interatomic potentials. J. Mol. Liq. 400, 124521 (2024)

    2024

  5. [13]

    Attarian, S., Morgan, D., Szlufarska, I.: Thermophysical properties of FLiBe using moment tensor potentials. J. Mol. Liq. 368, 120803 (2022) 19

    2022

  6. [14]

    Zhao, J., Feng, T., Lu, G.: Deep Learning Potential Assisted Prediction of Local Structure and Thermophysical Properties of the SrCl2-KCl-MgCl2 Melt. J. Chem. Theory Comput. (2024)

    2024

  7. [15]

    arXiv preprint arXiv:2401.00096 (2023)

    Batatia, I., Benner, P., Chiang, Y., Elena, A.M., Kov´ acs, D.P., Riebesell, J., Advincula, X.R., Asta, M., Baldwin, W.J., Bernstein, N., et al.: A foundation model for atomistic materials chemistry. arXiv preprint arXiv:2401.00096 (2023)

    2023 arXiv

  8. [16]

    Chen, C., Ong, S.P.: A universal graph deep learning interatomic potential for the periodic table. Nat. Comput. Sci. 2(11), 718–728 (2022)

    2022

  9. [17]

    Digital Discovery 2(2), 346–355 (2023)

    Choudhary, K., DeCost, B., Major, L., Butler, K., Thiyagalingam, J., Tavazza, F.: Unified graph neural network force-field for the periodic table: solid state applications. Digital Discovery 2(2), 346–355 (2023)

    2023

  10. [18]

    Nature Communications 13(1), 2991 (2022)

    Takamoto, S., Shinagawa, C., Motoki, D., Nakago, K., Li, W., Kurata, I., Watan- abe, T., Yayama, Y., Iriguchi, H., Asano, Y., et al.: Towards universal neural network potential for material discovery applicable to arbitrary combination of 45 elements. Nature Communications 13(...

    2022

  11. [19]

    arXiv preprint arXiv:2402.19327 (2024)

    Xie, F., Lu, T., Meng, S., Liu, M.: Gptff: A high-accuracy out-of-the-box universal ai force field for arbitrary inorganic materials. arXiv preprint arXiv:2402.19327 (2024)

    2024 arXiv

  12. [20]

    Nature 624(7990), 80–85 (2023)

    Merchant, A., Batzner, S., Schoenholz, S.S., Aykol, M., Cheon, G., Cubuk, E.D.: Scaling deep learning for materials discovery. Nature 624(7990), 80–85 (2023)

    2023

  13. [21]

    Park, Y., Kim, J., Hwang, S., Han, S.: Scalable parallel algorithm for graph neu- ral network interatomic potentials in molecular dynamics simulations. J. Chem. Theory Comput. (2024)

    2024

  14. [22]

    Deng, B., Zhong, P., Jun, K., Riebesell, J., Han, K., Bartel, C.J., Ceder, G.: CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling. Nat. Mach. Intell. 5(9), 1031–1041 (2023)

    2023

  15. [23]

    MGE Advances, 58 (2024)

    Yu, H., Giantomassi, M., Materzanini, G., Wang, J., Rignanese, G.-M.: Sys- tematic assessment of various universal machine-learning interatomic potentials. MGE Advances, 58 (2024)

    2024

  16. [24]

    Information science for materials discovery and design, 45–75 (2016)

    Frazier, P.I., Wang, J.: Bayesian optimization for materials design. Information science for materials discovery and design, 45–75 (2016)

    2016

  17. [25]

    Npj Comput

    Lookman, T., Balachandran, P.V., Xue, D., Yuan, R.: Active learning in materi- als science with emphasis on adaptive sampling using uncertainties for targeted design. Npj Comput. Mater. 5(1), 21 (2019)

    2019

  18. [26]

    Ng, K.W., Tian, G.-L., Tang, M.-L.: Dirichlet and related distributions: Theory, 20 methods and applications (2011)

    2011

  19. [27]

    Bochkarev, A., Lysogorskiy, Y., Ortner, C., Cs´ anyi, G., Drautz, R.: Multilayer atomic cluster expansion for semilocal interactions. Phys. Rev. Res. 4(4), 042019 (2022)

    2022

  20. [28]

    arXiv preprint arXiv:2312.15211 (2023)

    Kov´ acs, D.P., Moore, J.H., Browning, N.J., Batatia, I., Horton, J.T., Kapil, V., Witt, W.C., Magd˘ au, I.-B., Cole, D.J., Cs´ anyi, G.: MACE-OFF23: Trans- ferable machine learning force fields for organic molecules. arXiv preprint arXiv:2312.15211 (2023)

    2023 arXiv

  21. [29]

    In: Proceedings of the 25th International Conference on Machine Learning, pp

    Dasgupta, S., Hsu, D.: Hierarchical sampling for active learning. In: Proceedings of the 25th International Conference on Machine Learning, pp. 208–215 (2008)

    2008

  22. [30]

    Dasgupta, S.: Two faces of active learning. Theoret. Comput. Sci. 412(19), 1767– 1781 (2011)

    2011

  23. [31]

    In: Pacific-Asia Conference on Knowledge Discovery and Data Mining, pp

    Campello, R.J., Moulavi, D., Sander, J.: Density-based clustering based on hier- archical density estimates. In: Pacific-Asia Conference on Knowledge Discovery and Data Mining, pp. 160–172 (2013). Springer

    2013

  24. [32]

    McInnes, L., Healy, J., Astels, S., et al.: hdbscan: Hierarchical density based clustering. J. Open Source Softw. 2(11), 205 (2017)

    2017

  25. [33]

    In: 2017 IEEE International Conference on Data Mining Workshops (ICDMW), pp

    McInnes, L., Healy, J.: Accelerated hierarchical density based clustering. In: 2017 IEEE International Conference on Data Mining Workshops (ICDMW), pp. 33–42 (2017). IEEE

    2017

  26. [34]

    Melvin, R.L., Godwin, R.C., Xiao, J., Thompson, W.G., Berenhaut, K.S., Salsbury Jr, F.R.: Uncovering large-scale conformational change in molecular dynamics without prior knowledge. J. Chem. Theory Comput. 12(12), 6130–6146 (2016)

    2016

  27. [35]

    Cell Rep

    Li, Q.-J., K¨ u¸ c¨ ukbenli, E., Lam, S., Khaykovich, B., Kaxiras, E., Li, J.: Develop- ment of robust neural-network interatomic potential for molten salt. Cell Rep. Phys. Sci. 2(3) (2021)

    2021

  28. [36]

    CRC Handbook of Chemistry and Physics (2012)

    Haynes, W.: Density of molten elements and representative salts. CRC Handbook of Chemistry and Physics (2012)

    2012

  29. [37]

    Part III

    Wang, J., Wu, J., Lu, G., Yu, J.: Molecular dynamics study of the transport properties and local structures of molten alkali metal chlorides. Part III. Four binary systems LiCl-RbCl, LiCl-CsCl, NaCl-RbCl and NaCl-CsCl. J. Mol. Liq. 238, 236–247 (2017)

    2017

  30. [38]

    Solar Energy 152, 57–79 (2017)

    Li, Y., Xu, X., Wang, X., Li, P., Hao, Q., Xiao, B.: Survey and evaluation of equations for thermophysical properties of binary/ternary eutectic salts from nacl, 21 kcl, mgcl2, cacl2, zncl2 for heat transfer and thermal storage fluids in csp. Solar Energy 152, 57–79 (2017)

    2017

  31. [39]

    Nam, H., Bengtson, A., V¨ ortler, K., Saha, S., Sakidja, R., Morgan, D.: First- principles molecular dynamics modeling of the molten fluoride salt with Cr solute. J. Nucl. Mater. 449(1-3), 148–157 (2014)

    2014

  32. [40]

    Liu, M., Masset, P., Gray-Weale, A.: Solubility of sodium in sodium chloride: a density functional theory molecular dynamics study. J. Electrochem. Soc. 161(8), 3042 (2014)

    2014

  33. [41]

    Andersson, D., Beeler, B.W.: Ab initio molecular dynamics (AIMD) simulations of NaCl, UCl 3 and NaCl-UCl3 molten salts. J. Nucl. Mater. 568, 153836 (2022)

    2022

  34. [42]

    Advances in Neural Information Processing Systems 35, 11423–11436 (2022)

    Batatia, I., Kovacs, D.P., Simm, G., Ortner, C., Cs´ anyi, G.: Mace: Higher order equivariant message passing neural networks for fast and accurate force fields. Advances in Neural Information Processing Systems 35, 11423–11436 (2022)

    2022

  35. [43]

    Witt, W.C., Oord, C., Gelˇ zinyt˙ e, E., J¨ arvinen, T., Ross, A., Darby, J.P., Ho, C.H., Baldwin, W.J., Sachs, M., Kermode, J., et al.: ACE potentials: A Julia implementation of the atomic cluster expansion. J. Chem. Phys. 159(16) (2023)

    2023

  36. [44]

    arXiv preprint arXiv:2205.06643 (2022)

    Batatia, I., Batzner, S., Kov´ acs, D.P., Musaelian, A., Simm, G.N., Drautz, R., Ortner, C., Kozinsky, B., Cs´ anyi, G.: The design space of e(3)-equivariant atom- centered interatomic potentials. arXiv preprint arXiv:2205.06643 (2022)

    2022 arXiv

  37. [45]

    Darby, J.P., Kov´ acs, D.P., Batatia, I., Caro, M.A., Hart, G.L., Ortner, C., Cs´ anyi, G.: Tensor-reduced atomic density representations. Phys. Rev. Lett. 131(2), 028001 (2023)

    2023

  38. [46]

    Npj Comput

    Magd˘ au, I.-B., Arismendi-Arrieta, D.J., Smith, H.E., Grey, C.P., Hermansson, K., Cs´ anyi, G.: Machine learning force fields for molecular liquids: Ethylene Car- bonate/Ethyl Methyl Carbonate binary solvent. Npj Comput. Mater. 9(1), 146 (2023)

    2023

  39. [47]

    Kresse, G., Hafner, J.: Ab initio molecular dynamics for liquid metals. Phys. Rev. B 47(1), 558 (1993)

    1993

  40. [48]

    Perdew, J.P., Burke, K., Ernzerhof, M.: Generalized gradient approximation made simple. Phys. Rev. Lett. 77(18), 3865 (1996)

    1996

  41. [49]

    Bl¨ ochl, P.E.: Projector augmented-wave method. Phys. Rev. B 50(24), 17953 (1994)

    1994

  42. [50]

    Kresse, G., Joubert, D.: From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 59(3), 1758 (1999)

    1999

  43. [51]

    Grimme, S., Antony, J., Ehrlich, S., Krieg, H.: A consistent and accurate ab initio 22 parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 132(15) (2010)

    2010

  44. [52]

    Thompson, A.P., Aktulga, H.M., Berger, R., Bolintineanu, D.S., Brown, W.M., Crozier, P.S., In’t Veld, P.J., Kohlmeyer, A., Moore, S.G., Nguyen, T.D., et al.: LAMMPS-a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales. Comp...

    2022

  45. [53]

    Attarian, S., Shen, C., Morgan, D., Szlufarska, I.: Best practices for fitting machine learning interatomic potentials for molten salts: A case study using NaCl-MgCl2. Comput. Mater. Sci. 246, 113409 (2025)

    2025

  46. [54]

    Mart ´ ınez, L., Andrade, R., Birgin, E.G., Mart ´ ınez, J.M.: PACKMOL: A package for building initial configurations for molecular dynamics simulations. J. Comput. Chem. 30(13), 2157–2164 (2009)

    2009

  47. [55]

    Npj Comput

    Ward, L., Agrawal, A., Choudhary, A., Wolverton, C.: A general-purpose machine learning framework for predicting properties of inorganic materials. Npj Comput. Mater. 2(1), 1–7 (2016)

    2016

  48. [56]

    Ribeiro, M.H.D.M., Silva, R.G., Moreno, S.R., Mariani, V.C., Santos Coelho, L.: Efficient bootstrap stacking ensemble learning model applied to wind power generation forecasting. Int. J. Electr. Power Energy Syst. 136, 107712 (2022)

    2022

  49. [57]

    Journal of British Surgery 105(10), 1348–1348 (2018)

    Ranstam, J., Cook, J.A.: LASSO regression. Journal of British Surgery 105(10), 1348–1348 (2018)

    2018

  50. [58]

    Wiley Interdisciplinary Reviews: Computa- tional Statistics 1(1), 93–100 (2009)

    McDonald, G.C.: Ridge regression. Wiley Interdisciplinary Reviews: Computa- tional Statistics 1(1), 93–100 (2009)

    2009

  51. [59]

    Journal of Complexity 25(2), 201–230 (2009)

    De Mol, C., De Vito, E., Rosasco, L.: Elastic-net regularization in learning theory. Journal of Complexity 25(2), 201–230 (2009)

    2009

  52. [60]

    IEEE Access 9, 55999–56011 (2021)

    Luo, H., Cheng, F., Yu, H., Yi, Y.: SDTR: Soft decision tree regressor for tabular data. IEEE Access 9, 55999–56011 (2021)

    2021

  53. [61]

    Rigatti, S.J.: Random forest. J. Insur. Med. 47(1), 31–39 (2017)

    2017

  54. [62]

    Effi- cient learning machines: Theories, concepts, and applications for engineers and system designers, 67–80 (2015)

    Awad, M., Khanna, R., Awad, M., Khanna, R.: Support vector regression. Effi- cient learning machines: Theories, concepts, and applications for engineers and system designers, 67–80 (2015)

    2015

  55. [63]

    In: 2022 20th IEEE Interregional NEWCAS Conference (NEWCAS), pp

    Dutt, M.I., Saadeh, W.: A multilayer perceptron (MLP) regressor network for monitoring the depth of anesthesia. In: 2022 20th IEEE Interregional NEWCAS Conference (NEWCAS), pp. 251–255 (2022). IEEE

    2022

  56. [64]

    23 In: PyData 2014 (2014)

    Prettenhofer, P., Louppe, G.: Gradient boosted regression trees in scikit-learn. 23 In: PyData 2014 (2014)

    2014

  57. [65]

    Measurement: Sensors 35, 101277 (2024)

    Sudhamathi, T., Perumal, K.: Ensemble regression based extra tree regressor for hybrid crop yield prediction system. Measurement: Sensors 35, 101277 (2024)

    2024

  58. [66]

    In: Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining, pp

    Chen, T., Guestrin, C.: XGBoost: A scalable tree boosting system. In: Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining, pp. 785–794 (2016)

    2016

  59. [67]

    Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V.,et al.: Scikit-learn: Machine learning in Python. J. Mach. Learn. Res. 12, 2825–2830 (2011)

    2011

  60. [68]

    Proceedings of the IEEE 104(1), 148–175 (2015)

    Shahriari, B., Swersky, K., Wang, Z., Adams, R.P., De Freitas, N.: Taking the human out of the loop: A review of bayesian optimization. Proceedings of the IEEE 104(1), 148–175 (2015)

    2015

  61. [69]

    In: Practice and Experience in Advanced Research Computing, pp

    Brown, S.T., Buitrago, P., Hanna, E., Sanielevici, S., Scibek, R., Nystrom, N.A.: Bridges-2: A platform for rapidly-evolving and data intensive research. In: Practice and Experience in Advanced Research Computing, pp. 1–4 (2021)

    2021

  62. [70]

    In: Practice and Experience in Advanced Research Computing 2023: Computing for the Common Good, pp

    Boerner, T.J., Deems, S., Furlani, T.R., Knuth, S.L., Towns, J.: Access: Advanc- ing innovation: Nsf’s advanced cyberinfrastructure coordination ecosystem: Ser- vices & support. In: Practice and Experience in Advanced Research Computing 2023: Computing for the Common Good, pp....

    2023

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.