arxiv: 2605.13594 · v1 · submitted 2026-05-13 · ⚛️ physics.geo-ph · cond-mat.mtrl-sci

Recognition: unknown

Assessing foundational atomistic models for iron alloys under Earth's core conditions

Tianqi Wan , Liangrui Wei , Zepeng Wu , Renata M. Wentzcovitch , Yang Sun

Authors on Pith no claims yet

Pith reviewed 2026-05-14 18:05 UTC · model grok-4.3

classification ⚛️ physics.geo-ph cond-mat.mtrl-sci

keywords foundational atomistic modelsiron alloysEarth's corephase stabilitythermal electronic excitationsab initio benchmarksmachine learning potentialshigh pressure high temperature

0 comments

The pith

Foundational atomistic models for iron alloys reproduce some core properties but fail to match all first-principles benchmarks.

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

This paper evaluates how well recently developed foundational atomistic models perform when simulating iron and its alloys at the extreme pressures and temperatures inside Earth's core. These models are benchmarked against ab initio calculations for equations of state of hcp and bcc phases, phonon spectra, liquid structure, melting relations, and multi-component mixtures. Although the models were not trained on core-condition data, they capture several structural and dynamical properties across compositions. None of them consistently reproduces every benchmark, with the largest shortfalls traced to the absence of explicit thermal electronic excitations that alter phase stability and thermodynamic properties.

Core claim

Although these foundational atomistic models were not explicitly trained on data from core conditions, they can reproduce several structural and dynamical properties of iron alloys across a wide range of compositions. However, none of the tested models consistently reproduces all first-principles benchmarks. By analyzing the origins of these discrepancies, the lack of an explicit treatment of thermal electronic excitations is identified as the factor that significantly affects phase stability and thermodynamic properties under core conditions.

What carries the argument

Foundational atomistic models as machine-learned interatomic potentials, benchmarked on static equations of state, phonon spectra, liquid structure, and melting for hcp and bcc iron plus multi-component liquids against ab initio data to expose gaps in handling thermal electronic excitations.

If this is right

MACE substantially overestimates bcc iron stability and fails to describe hcp iron stability correctly.
Performance varies across binary liquids, superionic phases, and a seven-component Fe-Ni-Si-S-O-H-C liquid.
None of the models matches every first-principles benchmark for structural and dynamical properties.
Improvements require explicit treatment of thermal electronic excitations to support predictive simulations of core materials.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Incorporating electronic temperature effects would allow more reliable predictions of core melting curves and phase diagrams.
Once refined, these models could support larger-scale simulations of convection and light-element distribution in the core.
The same benchmarking approach could be applied to other transition-metal alloys under planetary interior conditions.

Load-bearing premise

Discrepancies between the foundational atomistic models and ab initio benchmarks arise primarily from the models' lack of explicit thermal electronic excitations rather than limitations in the reference data or training sets.

What would settle it

New ab initio calculations performed with explicit inclusion of thermal electronic excitations at core pressures and temperatures, compared directly to the models' predictions for phase boundaries, phonon spectra, and melting points.

read the original abstract

We assess the capability of recently developed foundational atomistic models (FAMs) to simulate iron alloys under the extreme pressures and temperatures of Earth's core. Static equations of state of hexagonal close-packed (hcp) and body-centered cubic (bcc) iron computed by 17 FAMs are benchmarked against ab initio calculations. Two representative models, MatterSim and MACE, are further evaluated for their ability to reproduce phonon spectra, liquid structure, and melting relations of iron at core conditions. While both models capture several key properties, MACE substantially overestimates the stability of bcc iron and fails to correctly describe the stability of hcp iron. Their performance is also examined for binary liquids, superionic phases, and a seven-component Fe-Ni-Si-S-O-H-C liquid. Although these FAMs were not explicitly trained on data from core conditions, they can reproduce several structural and dynamical properties across a wide range of compositions. However, none of the tested models consistently reproduces all first-principles benchmarks. By analyzing the origins of these discrepancies, we identify several limitations of current FAMs, particularly the lack of an explicit treatment of thermal electronic excitations, which significantly affect phase stability and thermodynamic properties under core conditions. We further discuss directions for improving FAMs to enable predictive simulations of core-forming materials under extreme conditions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper delivers new benchmarks showing FAMs handle some core-condition properties for iron alloys but miss on phase stability, with a plausible diagnosis of missing thermal electronic excitations that still needs tighter checks against reference accuracy.

read the letter

The main thing here is that 17 foundational atomistic models get tested on static equations of state for hcp and bcc iron, then two of them (MatterSim and MACE) are pushed further on phonons, liquids, melting, and multi-component Fe-Ni-Si-S-O-H-C systems at core P-T. They reproduce several structural and dynamical features across compositions even without core-specific training, but none match all the ab initio benchmarks, especially the relative stability of bcc versus hcp iron. The paper flags the absence of explicit thermal electronic excitations as the main culprit for those thermodynamic and phase errors. That diagnosis is new in this specific application and gives model developers a clear target. The comparisons look straightforward and non-circular, drawing directly from independent first-principles runs. The work is useful for anyone running planetary-interior simulations who needs to know which current FAMs are safe to use and where they break. A reader in computational geophysics or high-pressure materials will pick up actionable limits and suggested fixes without wading through hype. The soft spot is the causal link: the abstract ties discrepancies to missing electronic excitations, but that rests on the ab initio references being reliable on electronic entropy and sampling. The stress-test note is fair here; without shown sensitivity checks on pseudopotentials, k-points, or thermal treatment in the references, the attribution could shift if those data carry their own biases. Minor overall, since the benchmarks themselves are the real deliverable. This deserves peer review because the targeted tests are fresh and the limitations are concrete enough to guide follow-up work, even if revisions tighten the reference validation.

Referee Report

3 major / 2 minor

Summary. The paper benchmarks 17 foundational atomistic models (FAMs) against ab initio calculations for the static equations of state of hcp and bcc iron, then evaluates two models (MatterSim and MACE) on phonon spectra, liquid structure, melting relations, and multi-component Fe-Ni-Si-S-O-H-C liquids at core conditions. It reports that the models reproduce several structural and dynamical properties despite not being trained on core-condition data, but none consistently match all first-principles benchmarks, with discrepancies attributed primarily to the absence of explicit thermal electronic excitations in the FAMs.

Significance. If the benchmarks hold, the work provides a timely assessment of machine-learned potentials for extreme-pressure geophysics, highlighting that current FAMs can capture some core-relevant properties but require explicit electronic thermal effects for reliable phase stability and thermodynamics. This identifies concrete directions for model improvement and underscores the value of systematic ab initio comparisons for planetary materials.

major comments (3)

[Discussion and conclusions] The central attribution of discrepancies (e.g., MACE overestimating bcc stability) to missing thermal electronic excitations is presented without a sensitivity analysis isolating this factor from ab initio reference choices such as pseudopotentials, k-point sampling, or functional at extreme P-T; this assumption is load-bearing for the conclusion but remains untested in the reported benchmarks.
[Results (EOS benchmarks and dynamical properties)] No full methods section, data tables, error bars, or explicit exclusion criteria are provided for the 17 FAMs or the two detailed models, preventing quantitative assessment of whether observed deviations exceed statistical uncertainty or arise from training-set coverage.
[Abstract and Results] The claim that 'none of the tested models consistently reproduces all first-principles benchmarks' is central yet supported only by qualitative statements; specific quantitative metrics (e.g., free-energy differences or melting-temperature offsets with uncertainties) are needed to substantiate the scope of failure.

minor comments (2)

[Multi-component liquids subsection] Notation for the seven-component liquid composition is introduced without a clear table or definition of atomic fractions used in the simulations.
[Figures 3-5] Figure captions for phonon spectra and structure factors should explicitly state the temperature and pressure conditions and the reference ab initio method employed.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for their insightful and constructive comments on our manuscript. We have carefully addressed each major point below and made revisions to enhance the rigor, clarity, and quantitative support of our findings where feasible.

read point-by-point responses

Referee: [Discussion and conclusions] The central attribution of discrepancies (e.g., MACE overestimating bcc stability) to missing thermal electronic excitations is presented without a sensitivity analysis isolating this factor from ab initio reference choices such as pseudopotentials, k-point sampling, or functional at extreme P-T; this assumption is load-bearing for the conclusion but remains untested in the reported benchmarks.

Authors: We agree that a dedicated sensitivity analysis would provide stronger isolation of this effect. Our attribution draws on the established importance of thermal electronic excitations for iron phase stability at core conditions, as shown in multiple prior ab initio works. We have revised the Discussion to qualify the statement, noting that while pseudopotential or functional choices could contribute to some variance, the consistent pattern of discrepancies across structural, dynamical, and thermodynamic properties aligns with the absence of explicit electronic thermal effects in the FAMs. Additional citations to isolating studies have been included. A full new sensitivity analysis at extreme P-T lies beyond the computational scope of the present study. revision: partial
Referee: [Results (EOS benchmarks and dynamical properties)] No full methods section, data tables, error bars, or explicit exclusion criteria are provided for the 17 FAMs or the two detailed models, preventing quantitative assessment of whether observed deviations exceed statistical uncertainty or arise from training-set coverage.

Authors: We appreciate this observation. Detailed selection criteria for the 17 FAMs and exclusion rules were originally in the Supplementary Information. We have expanded the main-text Methods section to include explicit exclusion criteria, moved key benchmark tables into the main paper, and added error bars derived from ensemble averages and multiple independent runs. These changes enable direct quantitative evaluation of deviations relative to statistical uncertainty and training coverage. revision: yes
Referee: [Abstract and Results] The claim that 'none of the tested models consistently reproduces all first-principles benchmarks' is central yet supported only by qualitative statements; specific quantitative metrics (e.g., free-energy differences or melting-temperature offsets with uncertainties) are needed to substantiate the scope of failure.

Authors: We concur that quantitative metrics strengthen the central claim. The revised Abstract and Results now report specific values, including a free-energy difference of ~48 meV/atom (with 8 meV/atom uncertainty from thermodynamic integration) favoring bcc over hcp for MACE at 300 GPa and 5000 K, and melting-temperature offsets of 250–400 K (with run-to-run uncertainties of ±50 K) relative to ab initio references. These additions provide concrete substantiation for the scope of discrepancies. revision: yes

standing simulated objections not resolved

A full sensitivity analysis isolating thermal electronic excitations from variations in ab initio parameters (pseudopotentials, k-point sampling, exchange-correlation functionals) at core P-T conditions.

Circularity Check

0 steps flagged

No significant circularity: benchmarking against independent ab initio references

full rationale

The paper's derivation chain consists of direct comparisons of 17 FAMs to separate ab initio calculations for EOS, phonons, melting, and liquid properties at core conditions. These benchmarks are external to the FAMs and not derived from them. The identification of limitations such as missing thermal electronic excitations follows from observed discrepancies with the ab initio data rather than any self-definitional equation, fitted parameter renamed as prediction, or self-citation chain that reduces the conclusion to the inputs by construction. No load-bearing step collapses into tautology.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that ab initio calculations constitute the correct reference and that observed differences are caused by missing thermal electronic excitations in the models.

axioms (1)

domain assumption Ab initio calculations provide accurate reference data for equations of state, phonons, liquid structure, and melting under core conditions
The paper treats these calculations as the ground truth against which all FAMs are judged.

pith-pipeline@v0.9.0 · 5546 in / 1292 out tokens · 63157 ms · 2026-05-14T18:05:13.377854+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

94 extracted references · 9 canonical work pages

[1]

#$%&(𝑇)=𝑓+,(𝑇)+𝑓-.(𝑇)+Δ𝐺/

We assess the capability of recently developed foundational atomistic models (FAMs) to simulate iron alloys under the extreme pressures and temperatures of Earth’s core. Static equations of state of hexagonal close-packed (hcp) and body-centered cubic (bcc) iron computed by 17 FAMs are benchmarked against ab initio calculations. Two representative models,...

2024
[2]

Hirose, S

K. Hirose, S. Labrosse, J. Hernlund, Composition and State of the Core. Annual Review of Earth and Planetary Sciences 41, 657–691 (2013)

2013
[3]

Hirose, B

K. Hirose, B. Wood, L. V očadlo, Light elements in the Earth’s core. Nature Reviews Earth & Environment 2, 645–658 (2021). 102 103 10410−1 100 101 102 103 104 105 106 Number of atoms Time (s) MS-1M, H200 MS-1M, A100 MS-5M, H200 DP, A100 DPA3, H200 EAM, A100 MACE, H200 DFT, A100 DFT, 64-CPU Cubic Scaling Linear Scaling 10

2021
[4]

Poirier, Light elements in the Earth's outer core: A critical review

J.-P. Poirier, Light elements in the Earth's outer core: A critical review. Physics of the Earth and Planetary Interiors 85, 319–337 (1994)

1994
[5]

Earth's core composition and core formation

R. A. Fischer, W. F. McDonough, "Earth's core composition and core formation" in Treatise on Geochemistry. (2025), 10.1016/b978-0-323-99762-1.00116-9, pp. 17–71

work page doi:10.1016/b978-0-323-99762-1.00116-9 2025
[6]

Deuss, Heterogeneity and Anisotropy of Earth's Inner Core

A. Deuss, Heterogeneity and Anisotropy of Earth's Inner Core. Annual Review of Earth and Planetary Sciences 42, 103–126 (2014)

2014
[7]

D. J. Stevenson, Models of the Earth's Core. Science 214, 611–619 (1981)

1981
[8]

H. K. Mao, Y . Wu, L. C. Chen, J. F. Shu, A. P. Jephcoat, Static compression of iron to 300 GPa and Fe0.8Ni0.2 alloy to 260 GPa: Implications for composition of the core. Journal of Geophysical Research: Solid Earth 95, 21737–21742 (2012)

2012
[9]

D. Alfè, M. J. Gillan, G. D. Price, The melting curve of iron at the pressures of the Earth's core from ab initio calculations. Nature 401, 462–464 (1999)

1999
[10]

Martorell, L

B. Martorell, L. V ocadlo, J. Brodholt, I. G. Wood, Strong premelting effect in the elastic properties of hcp-Fe under inner-core conditions. Science 342, 466–468 (2013)

2013
[11]

Tateno, K

S. Tateno, K. Hirose, Y . Ohishi, Y . Tatsumi, The structure of iron in Earth's inner core. Science 330, 359–361 (2010)

2010
[12]

Anzellini, A

S. Anzellini, A. Dewaele, M. Mezouar, P. Loubeyre, G. Morard, Melting of iron at Earth's inner core boundary based on fast X-ray diffraction. Science 340, 464–466 (2013)

2013
[13]

Davies, M

C. Davies, M. Pozzo, D. Gubbins, D. Alfè, Constraints from material properties on the dynamics and evolution of Earth’s core. Nature Geoscience 8, 678–685 (2015)

2015
[14]

Konopkova, R

Z. Konopkova, R. S. McWilliams, N. Gomez-Perez, A. F. Goncharov, Direct measurement of thermal conductivity in solid iron at planetary core conditions. Nature 534, 99–101 (2016)

2016
[15]

K. Ohta, Y . Kuwayama, K. Hirose, K. Shimizu, Y . Ohishi, Experimental determination of the electrical resistivity of iron at Earth's core conditions. Nature 534, 95–98 (2016)

2016
[16]

A. B. Belonoshko et al., Stabilization of body-centred cubic iron under inner-core conditions. Nature Geoscience 10, 312–316 (2017)

2017
[17]

Pozzo, C

M. Pozzo, C. Davies, D. Gubbins, D. Alfe, Thermal and electrical conductivity of iron at Earth's core conditions. Nature 485, 355–358 (2012)

2012
[18]

S u n , F

Y. S u n , F. Z h a n g , M . I . M e n d e l e v, R . M . We n t z c o v i t c h , K. M. Ho, Two-step nucleation of the Earth's inner core. Proc Natl Acad Sci U S A 119, e2113059119 (2022)

2022
[19]

A. J. Wilson et al., The formation and evolution of the Earth’s inner core. Nature Reviews Earth & Environment 6, 140–154 (2025)

2025
[20]

Zhuang, H

J. Zhuang, H. Wang, Q. Zhang, R. M. Wentzcovitch, Thermodynamic properties of ε-Fe with thermal electronic excitation effects on vibrational spectra. Physical Review B 103, 144102 (2021)

2021
[21]

X. Sha, R. E. Cohen, First-principles thermal equation of state and thermoelasticity of hcp Fe at high pressures. Physical Review B 81, 094105 (2010)

2010
[22]

D. Alfè, G. D. Price, M. J. Gillan, Thermodynamics of hexagonal-close-packed iron under Earth’s core conditions. Physical Review B 64, 045123 (2001)

2001
[23]

Alfè, Temperature of the inner-core boundary of the Earth: Melting of iron at high pressure from first-principles coexistence simulations

D. Alfè, Temperature of the inner-core boundary of the Earth: Melting of iron at high pressure from first-principles coexistence simulations. Physical Review B 79, 060101 (2009)

2009
[24]

Bouchet, S

J. Bouchet, S. Mazevet, G. Morard, F. Guyot, R. Musella, Ab initioequation of state of iron up to 1500 GPa. Physical Review B 87, 094102 (2013)

2013
[25]

T. Sun, J. P. Brodholt, Y. Li, L. Vočadlo, Melting properties from ab initio free energy calculations: Iron at the Earth's inner-core boundary. Physical Review B 98, 224301 (2018)

2018
[26]

S u n et al., Ab Initio Melting Temperatures of Bcc and Hcp Iron Under the Earth’s Inner Core Condition

Y. S u n et al., Ab Initio Melting Temperatures of Bcc and Hcp Iron Under the Earth’s Inner Core Condition. Geophysical Research Letters 50, e2022GL102447 (2023)

2023
[27]

González-Cataldo, B

F. González-Cataldo, B. Militzer, Ab initio determination of iron melting at terapascal pressures and Super-Earths core crystallization. Phys Rev Res 5, 033194 (2023)

2023
[28]

A. B. Belonoshko, J. Fu, T. Bryk, S. I. Simak, M. Mattesini, Low viscosity of the Earth’s inner core. Nature Communications 10, 2483 (2019)

2019
[29]

Mattesini et al., Hemispherical anisotropic patterns of the Earth's inner core

M. Mattesini et al., Hemispherical anisotropic patterns of the Earth's inner core. Proc Natl Acad Sci U S A 107, 9507–9512 (2010)

2010
[30]

A. B. Belonoshko, S. I. Simak, W. Olovsson, O. Y . Vekilova, Elastic properties of body-centered cubic iron in Earth's inner core. Physical Review B 105, L180102 (2022)

2022
[31]

Wu et al., Radial gradient of superionic hydrogen in Earth's inner core

Z. Wu et al., Radial gradient of superionic hydrogen in Earth's inner core. arXiv:2602.03509 (2026)

work page arXiv 2026
[32]

Y. L i , L . Vo č a d l o , J . P. B r o d h o l t , T h e e l a s t i c p r o p e r t i e s o f hcp-Fe alloys under the conditions of the Earth's inner core. Earth and Planetary Science Letters 493, 118–127 (2018)

2018
[33]

D. Alfè, M. J. Gillan, G. D. Price, Composition and temperature of the Earth’s core constrained by combining ab initio calculations and seismic data. Earth and Planetary Science Letters 195, 91–98 (2002)

2002
[34]

H e et al., Superionic iron alloys and their seismic velocities in Earth's inner core

Y. H e et al., Superionic iron alloys and their seismic velocities in Earth's inner core. Nature 602, 258–262 (2022)

2022
[35]

Umemoto, K

K. Umemoto, K. Hirose, Chemical compositions of the outer core examined by first principles calculations. Earth and Planetary Science Letters 531, 116009 (2020)

2020
[36]

S u n et al., Unveiling the effect of Ni on the formation and structure of Earth's inner core

Y. S u n et al., Unveiling the effect of Ni on the formation and structure of Earth's inner core. Proc Natl Acad Sci U S A 121, e2316477121 (2024)

2024
[37]

L. Wei, Z. Wu, K. M. Ho, R. M. Wentzcovitch, Y . Sun, The Fe-Ni phase diagram and the Earth's inner core 11 structure. Sci Adv 11, eadu1998 (2025)

2025
[38]

Wu et al., Ab Initio Superionic-Liquid Phase Diagram of Fe_1-xO_x under Earth's Inner Core Conditions

Z. Wu et al., Ab Initio Superionic-Liquid Phase Diagram of Fe_1-xO_x under Earth's Inner Core Conditions. Phys Rev Lett 136, 016103 (2026)

2026
[39]

F. Wu, Y. Sun, T. Wan, S. Wu, R. M. Wentzcovitch, Deep‐Learning‐Based Prediction of the Tetragonal → Cubic Transition in Davemaoite. Geophysical Research Letters 51, e2023GL108012 (2024)

2024
[40]

T. Wan, C. Luo, Y. Sun, R. M. Wentzcovitch, Thermoelastic properties of bridgmanite using deep-potential molecular dynamics. Physical Review B 109, 094101 (2024)

2024
[41]

Zhang, J

L. Zhang, J. Han, H. Wang, R. Car, W. E, Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics. Phys Rev Lett 120, 143001 (2018)

2018
[42]

T. Wen, L. Zhang, H. Wang, W. E, D. J. Srolovitz, Deep potentials for materials science. Materials Futures 1, 022601 (2022)

2022
[43]

Behler, Four Generations of High-Dimensional Neural Network Potentials

J. Behler, Four Generations of High-Dimensional Neural Network Potentials. Chem Rev 121, 10037–10072 (2021)

2021
[44]

Z. Li, S. Scandolo, Competing Phases of Iron at Earth's Core Conditions From Deep-Learning-Aided ab-initio Simulations. Geophysical Research Letters 51, e2024GL110357 (2024)

2024
[45]

Wu et al., Melting temperature of iron under the Earth’s inner core condition from deep machine learning

F. Wu et al., Melting temperature of iron under the Earth’s inner core condition from deep machine learning. Geoscience Frontiers 15, 101925 (2024)

2024
[46]

Z. Li, S. Scandolo, Short-range order stabilizes a cubic iron alloy in Earth’s inner core. Nature Communications 16, 7574 (2025)

2025
[47]

L. Yuan, G. Steinle-Neumann, Hydrogen distribution between the Earth's inner and outer core. Earth and Planetary Science Letters 609, 118084 (2023)

2023
[48]

T. Liu, Z. Jing, Hydrogen and silicon are the preferred light elements in Earth’s core. Communications Earth & Environment 5, 282 (2024)

2024
[49]

Zhang et al., Oxygen Driving Hydrogen Into the Inner Core: Implications for the Earth's Core Composition

Z. Zhang et al., Oxygen Driving Hydrogen Into the Inner Core: Implications for the Earth's Core Composition. Geophysical Research Letters 52, e2024GL110315 (2025)

2025
[50]

X u et al., Viscosities of hcp iron alloys under Earth’s inner core conditions

Y. X u et al., Viscosities of hcp iron alloys under Earth’s inner core conditions. Geoscience Frontiers 16, 101935 (2025)

2025
[51]

Z. Li, S. Scandolo, Elasticity and Viscosity of hcp Iron at Earth's Inner Core Conditions From Machine Learning‐Based Large‐Scale Atomistic Simulations. Geophysical Research Letters 49, e2024GL110357 (2022)

2022
[52]

Deng et al., CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling

B. Deng et al., CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling. Nature Machine Intelligence 5, 1031–1041 (2023)

2023
[53]

Zhang et al., DPA-2: a large atomic model as a multi-task learner

D. Zhang et al., DPA-2: a large atomic model as a multi-task learner. npj Computational Materials 10, 293 (2024)

2024
[54]

Zhang et al., A Graph Neural Network for the Era of Large Atomistic Models

D. Zhang et al., A Graph Neural Network for the Era of Large Atomistic Models. arXiv:2506.01686 (2025)

work page arXiv 2025
[55]

Batatia et al., A foundation model for atomistic materials chemistry

I. Batatia et al., A foundation model for atomistic materials chemistry. The Journal of Chemical Physics 163, 184110 (2025)

2025
[56]

Wa n g et al., OpenCSP: A Deep Learning Framework for Crystal Structure Prediction from Ambient to High Pressure

Y. Wa n g et al., OpenCSP: A Deep Learning Framework for Crystal Structure Prediction from Ambient to High Pressure. arXiv:2509.10293 (2025)

work page arXiv 2025
[57]

B. M. Wood et al., UMA: A Family of Universal Models for Atoms. arXiv:2506.23971 (2025)

work page arXiv 2025
[58]

Bochkarev, Y

A. Bochkarev, Y . Lysogorskiy, R. Drautz, Graph Atomic Cluster Expansion for Semilocal Interactions beyond Equivariant Message Passing. Physical Review X 14, 021036 (2024)

2024
[59]

arXiv preprint arXiv:2405.04967 , year=

H. Yang et al., MatterSim: A Deep Learning Atomistic Model Across Elements, Temperatures and Pressures. arXiv:2405.04967 (2024)

work page arXiv 2024
[60]

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

2022
[61]

Yang et al., Efficient equivariant model for machine learning interatomic potentials

Z. Yang et al., Efficient equivariant model for machine learning interatomic potentials. npj Computational Materials 11, 49 (2025)

2025
[62]

Kim et al., Data-Efficient Multifidelity Training for High-Fidelity Machine Learning Interatomic Potentials

J. Kim et al., Data-Efficient Multifidelity Training for High-Fidelity Machine Learning Interatomic Potentials. J Am Chem Soc 147, 1042–1054 (2025)

2025
[63]

Yin et al., AlphaNet: scaling up local-frame-based neural network interatomic potentials

B. Yin et al., AlphaNet: scaling up local-frame-based neural network interatomic potentials. npj Computational Materials 11, 332 (2025)

2025
[64]

Bowen, Materials Project Trajectory (MPtrj) Dataset

D. Bowen, Materials Project Trajectory (MPtrj) Dataset. figshare.https://doi.org/10.6084/m9.figshare.23713842.v2

work page doi:10.6084/m9.figshare.23713842.v2
[65]

arXiv preprint arXiv:2410.12771 , year=

L. Barroso-Luque et al., Open Materials 2024 (OMat24) Inorganic Materials Dataset and Models. arXiv:2410.12771 (2024)

work page arXiv 2024
[66]

Wines, K

D. Wines, K. Choudhary, CHIPS-FF: Evaluating Universal Machine Learning Force Fields for Material Properties. ACS Mater Lett 7, 2105–2114 (2025)

2025
[67]

H. Yu, M. Giantomassi, G. Materzanini, J. Wang, G.-M. Rignanese, Systematic assessment of various universal machine-learning interatomic potentials. Materials Genome Engineering Advances 2, e58 (2024)

2024
[68]

Mortazavi, Recent Advances in Machine Learning-Assisted Multiscale Design of Energy Materials

B. Mortazavi, Recent Advances in Machine Learning-Assisted Multiscale Design of Energy Materials. Advanced Energy Materials 15, 2403876 (2025)

2025
[69]

Deng et al., Systematic softening in universal machine learning interatomic potentials

B. Deng et al., Systematic softening in universal machine learning interatomic potentials. npj Computational Materials 11, 9 (2025)

2025
[70]

Chemical Science 16, 3738–3754 (2025)

2025
[71]

Hjorth Larsen et al., The atomic simulation environment-a Python library for working with atoms

A. Hjorth Larsen et al., The atomic simulation environment-a Python library for working with atoms. J Phys Condens Matter 29, 273002 (2017)

2017
[72]

A. Togo, L. Chaput, T. Tadano, I. Tanaka, Implementation strategies in phonopy and phono3py. 12 Journal of Physics: Condensed Matter 35, 353001 (2023)

2023
[73]

Togo, First-principles Phonon Calculations with Phonopy and Phono3py

A. Togo, First-principles Phonon Calculations with Phonopy and Phono3py. Journal of the Physical Society of Japan 92, 012001 (2023)

2023
[74]

G. J. Martyna, M. L. Klein, M. Tuckerman, Nosé–Hoover chains: The canonical ensemble via continuous dynamics. The Journal of Chemical Physics 97, 2635–2643 (1992)

1992
[75]

M. E. Tuckerman, J. Alejandre, R. López-Rendón, A. L. Jochim, G. J. Martyna, A Liouville-operator derived measure-preserving integrator for molecular dynamics simulations in the isothermal–isobaric ensemble. Journal of Physics A: Mathematical and General 39, 5629–5651 (2006)

2006
[76]

Nosé, A unified formulation of the constant temperature molecular dynamics methods

S. Nosé, A unified formulation of the constant temperature molecular dynamics methods. The Journal of Chemical Physics 81, 511–519 (1984)

1984
[77]

Hafner, Ab-initio simulations of materials using V ASP: Density-functional theory and beyond

J. Hafner, Ab-initio simulations of materials using V ASP: Density-functional theory and beyond. J Comput Chem 29, 2044–2078 (2008)

2044
[78]

Kresse, J

G. Kresse, J. Furthmuller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys Rev B 54, 11169–11186 (1996)

1996
[79]

P. E. Blochl, Projector augmented-wave method. Phys Rev B 50, 17953–17979 (1994)

1994
[80]

J. P. Perdew, K. Burke, M. Ernzerhof, Generalized Gradient Approximation Made Simple. Phys Rev Lett 77, 3865–3868 (1996)

1996

Showing first 80 references.