Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Predicting neutron experiments from first principles: A workflow powered by machine learning

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Simulated neutron spectra from first-principles dynamics match measurements on four spectrometers.

desk verdict A useful, reproducible integration of MLIP-based MD with instrument-specific INS prediction, held back by one partly circular validation choice and overstated conclusions. read the letter →

arxiv 2504.19352 v1 pith:ONS32TTP submitted 2025-04-27 cond-mat.mtrl-sci cond-mat.softphysics.comp-ph

Eric Lindgren , Adam J. Jackson , Erik Fransson , Esmée Berger , Svemir Rudić , Goran Škoro , Rastislav Turanyi , Sanghamitra Mukhopadhyay
show 1 more author
Paul Erhart
This is my paper · ORCID
classification cond-mat.mtrl-scicond-mat.softphysics.comp-ph
keywords inelasticneutronscatteringmachine-learnedinteratomicpotentialsmoleculardynamicsdynamicstructurefactorquantumcorrectioninstrumentresolutionfunctionkinematicconstraintanharmonicity
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

This paper establishes that a purely computational workflow can predict neutron scattering experiments closely enough to guide and interpret them. The chain starts with density functional theory data, trains machine-learned interatomic potentials, runs large molecular dynamics simulations, and converts the trajectories into instrument-specific inelastic neutron scattering spectra. Tested on crystalline silicon, crystalline benzene, and hydrogenated scandium-doped barium titanate, the simulated spectra reproduce nearly every vibrational peak seen on four different neutron spectrometers. The result matters because it gives materials scientists a way to plan, validate, and interpret scattering experiments without waiting for beamline access.

What carries the argument

The load-bearing object is the dynamic structure factor $S(\mathbf{q},\omega)$, the time Fourier transform of the intermediate scattering function $F(\mathbf{q},t)$, which is the autocorrelation of the Fourier-transformed particle density. $S(\mathbf{q},\omega)$ is proportional to the intensity measured in a scattering experiment, and molecular dynamics supplies it directly, including anharmonic and multi-phonon content that harmonic phonon models must add by hand. What turns a generic spectrum into a prediction is the instrument-specific layer: a kinematic mask from the time-of-flight geometry, an energy-dependent Gaussian resolution function, and the first-order Stokes-Raman quantum correction $\beta\hbar\omega/(1-e^{-\beta\hbar\omega})$ applied to correct for classical phonon statistics. Each component is needed; the benzene comparison shows the raw spectrum improves markedly only after both the kinematic constraint and the quantum correction are applied.

What would settle it

Measure the absolute intensity of a clean, isolated overtone band in a crystal and compare it with the workflow's prediction: the first-order quantum correction fixes its intensity, so a factor-of-two disagreement would falsify that correction layer. The paper already shows this test partially failing in hydrogenated scandium-doped barium titanate, where the 250 meV overtone is underestimated and the 550 meV combination mode is absent.

Watch

Extended reading notes

Core claim

The paper's central claim is that inelastic neutron scattering spectra can be computed from first principles with quantitative fidelity, provided every step of the simulation is made to speak the instrument's language. Molecular dynamics trajectories from machine-learned potentials are transformed into the dynamic structure factor $S(\mathbf{q},\omega)$; the result is weighted by species-dependent neutron scattering lengths, masked by the spectrometer's kinematic constraint, convolved with an energy-dependent Gaussian resolution function, and multiplied by a first-order quantum correction factor that restores Bose statistics to the classically occupied phonon modes. Across silicon, benzene, and hydrogenated scandium-doped barium titanate, this procedure reproduces essentially all experimental vibrational peaks, including anharmonic mode softening with temperature, multi-phonon intensity between acoustic and optical branches, and the spectral differences between competing crystal phases. The paper states the agreement as quantitative and notes that almost all experimental features are faithfully reproduced.

Load-bearing premise

The load-bearing premise is that simplified corrections—a one-dimensional Gaussian resolution function and a first-order quantum factor—accurately convert classical molecular dynamics spectra into measured neutron intensities; the paper itself notes the true resolution is four-dimensional and non-Gaussian and that higher-order scattering processes require corrections this workflow does not apply.

Editorial extensions

If this is right

  • Simulated spectra can be used to identify and assign measured vibrational modes, since the workflow yields the spectrum and the underlying phonon dispersion from the same potential.
  • The method differentiates crystal phases by comparing simulated spectra of each phase against experiment, demonstrated by distinguishing hexagonal from cubic scandium-doped barium titanate as doping varies.
  • Anharmonicity, thermal expansion, and multi-phonon effects are included inherently, so the workflow predicts mode softening and finite-temperature spectral changes without perturbative corrections.
  • The same dynamic-structure-factor pipeline extends to diffraction and to X-ray or electron scattering by changing the weighting factors, as the paper argues.
  • The remaining quantitative gap, underestimated overtone and missing combination-mode intensities, can be reduced by higher-order quantum corrections, although the paper recommends the lowest-order correction for general guidance.

Reading between the lines

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

  • A natural next stress test is a disordered or liquid system, where the same dynamic structure factor pipeline should capture diffusive quasi-elastic broadening; the paper asserts applicability but does not demonstrate it.
  • If the workflow is adopted at user facilities, simulated instrument responses could be inverted against desired scientific outcomes to optimize which spectrometer, energy range, and counting time a proposal should request.
  • The high-order combination-mode failure suggests a targeted extension: apply higher-order quantum corrections only in energy windows where an independent harmonic analysis identifies multi-phonon contributions, leaving first-order regions unchanged.
  • General-purpose machine-learned potentials could replace bespoke training for new materials, turning the workflow into a nearly turnkey predictive tool at the cost of heavier per-simulation compute.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 a computational workflow for predicting inelastic neutron scattering (INS) spectra from first principles. The workflow combines density functional theory, neuroevolution-potential (NEP) machine-learned interatomic potentials, molecular dynamics (including path-integral MD), autocorrelation analysis via dynasor, instrument-specific resolution functions and kinematic constraints, and a first-order quantum correction factor. The authors demonstrate the workflow on three systems: crystalline silicon at several temperatures compared with ARCS data, crystalline benzene at 127 K compared with TOSCA data, and hydrogenated Sc-doped BaTiO3 at various Sc concentrations compared with IN1 Lagrange, TOSCA, and MAPS data. The paper reports good agreement in peak positions and many relative intensities, while also identifying limitations for higher-order multi-phonon features and systematic shifts attributed to the underlying DFT functionals.

Significance. If the validation concerns are addressed, this is a valuable methods contribution. The workflow is carefully assembled from open-source components, the models and training data are made available on Zenodo, and the central pipeline from MD trajectories to instrument-specific INS predictions is plausible and reproducible. The paper does not reduce any derived equation to a fitted parameter; the free parameters are limited to normalization/scaling choices and the selection of a specific Brillouin zone in one comparison. The demonstrations span three chemically different systems and four instruments, which gives the workflow breadth. The explicit recognition of the limitations of the first-order quantum correction and the comparison with harmonic AbINS calculations is a strength. The main weakness is that the strongest validation claim, particularly for temperature-dependent anharmonicity in silicon, relies on a partly circular selection procedure that must be repaired before the predictive claim can be accepted.

major comments (3)
  1. [III A, Fig. 2c and Fig. 4] The 1500 K silicon validation is partly circular and therefore cannot support the claim that the workflow captures anharmonicity and multi-phonon effects as a function of temperature. The text states that the Brillouin zone shown in Fig. 2c was selected because it minimizes the mean-squared error against the experimental spectrum, and Fig. 4a shows that the multi-phonon intensity near 30 meV varies substantially among the 52 zones. With this selection rule, the displayed curve is not an independent prediction; it is the best-fitting zone chosen using the target data. Please report the full distribution over all 52 zones, for example as an ensemble band or as an average with spread, and draw the validation conclusion from the distribution without using the experimental spectrum to select the displayed simulation.
  2. [II C, Eq. (6); III C; III D] The paper acknowledges that the quantum correction in Eq. (6) is valid only for first-order Stokes scattering, and Section III C shows the consequence: the 250 meV overtone is underestimated and the 550 meV combination mode is absent from the MD workflow. Given these results, the Discussion's statement that 'almost all experimental features . . . are faithfully reproduced' and that relative intensities are reproduced 'after applying correction factors for quantum statistics' overstates what is demonstrated. Please restrict the validation claim to first-order features, state explicitly that higher-order multi-phonon intensities are not yet predictive in the MD-based workflow, and temper the wording in Section III D accordingly.
  3. [III B, Fig. 5] The benzene comparison uses per-spectrum intensity rescaling: the caption states that the simulated spectra are individually scaled to match the experiment above 50 meV. Because the relative intensity between high- and low-energy regions is part of the quantitative validation, please clarify whether this is a single fixed scale factor for each spectrum or a free normalization, and report how the scaling affects the comparison. The current description makes it difficult to judge how much of the apparent agreement comes from this adjustable normalization.
minor comments (5)
  1. [II, first paragraph] The sentence 'The first step is the the construction' contains a duplicated article; please correct it.
  2. [Fig. 5(b)] The axis label 'Freuency' should be 'Frequency'.
  3. [Fig. 7] The shared color scale clips the intensity in panel (a); consider using per-panel normalization or an unclipped supplementary version so that the fundamental-mode intensity is visible.
  4. [Fig. 2 caption] The phrase 'the dispersion in (a) and intensity in (b) is aggregated' has a subject-verb disagreement; it should read 'are aggregated'.
  5. [VI, Data Availability] The data availability statement gives Zenodo links for the benzene and Sc-doped BaTiO3 models but not for the silicon NEP model; since the silicon model is taken from Ref. [17], please state explicitly where that model can be obtained.

Circularity Check

1 steps flagged · score 6.0 of 10

Post-hoc selection of the best-matching Brillouin zone at 1500 K makes the Si temperature-series validation circular; the rest of the workflow is otherwise self-contained.

  1. fitted input called prediction [Section III A (Si anharmonicity), discussion of Fig. 2c and Fig. 4a]
    "In fact, the intensity at the X-point varies substantially depending on the Brillouin zone, especially the intensity of the multi-phonon shoulder around 30 meV (Fig. 4). For the comparison shown in (Fig. 2c), we selected the Brillouin zone for which the simulated spectrum best reproduces the experimental data, based on the mean-squared error calculated over the spectrum."

    The target neutron spectrum is used as the selection criterion among the 52 simulated S(q,omega) curves for the 1500 K X-point comparison. Since Fig. 4a shows that the multi-phonon shoulder near 30 meV varies strongly with Brillouin zone, the curve displayed in Fig. 2c is, by construction, the one with the smallest mean-squared error to the experimental data; its agreement with experiment is therefore an optimized postdiction rather than an independent prediction.

full rationale

The remainder of the derivation chain is self-contained. The Si 300 K comparison aggregates all 52 Brillouin zones before comparison, so no experimental information enters the choice of displayed spectrum; the benzene and Sc-doped BaTiO3 simulations use pre-specified compositions, phases, temperatures, and instrument masks, and their residual discrepancies are attributed to the DFT functional or the first-order quantum correction rather than to any parameter fitted on the target spectra. The machine-learned potentials are trained on DFT energies, forces, and virials, not on the experimental INS data; the dynasor, euphonic, ResINS/Mantid, and PyChop components are independent published tools; and the quantum-correction factor of Eq. (6) is a standard first-order Stokes factor with stated validity limits that the paper itself tests against the 250 meV overtone and 550 meV combination band. No equation in the paper equals its input by construction, and no central uniqueness theorem is imported from the authors' prior work. The only genuine circular step is the post-hoc Brillouin-zone selection in Section III A. Because that selection is used as load-bearing validation for the temperature-dependent anharmonicity claim, the paper's overall statement of remarkable quantitative agreement is partially circular, though the workflow as a whole retains substantial independent content.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claim rests on several domain assumptions, most of which the authors state explicitly. No invented physical entities are introduced. The main free parameters are a per-spectrum intensity scaling and the post-hoc Brillouin zone choice for the 1500 K silicon comparison; both reduce the strength of the validation but do not invalidate the workflow.

free parameters (3)
  • Per-spectrum intensity scaling factor = Not specified (each simulated spectrum scaled individually)
    Used for benzene and implicitly for other spectra to compare shapes; preserves relative peak intensities within a spectrum but means absolute scattered intensity is not predicted.
  • Brillouin zone index for 1500 K silicon X-point spectrum = q = [0.5,-1.0,-1.5] (third Brillouin zone)
    Selected because it produced the smallest mean-squared error against the experimental spectrum out of 52 sampled zones; this is a discrete parameter chosen to match the target.
  • Gaussian broadening width for q-point sampling = 0.01 inverse Angstrom
    Applied to each q-point before spherical-shell averaging; chosen by hand with no sensitivity analysis.
assumptions (6)
  • domain assumption DFT with vdW-DF-cx for benzene and r2SCAN for Sc-doped BaTiO3 provides reference energies, forces, and stresses accurate enough for the spectra to match experiment.
    All MLIP training data come from these DFT calculations (Section II D). The authors attribute the 10-25 meV shifts in peak positions to the DFT functional, so this assumption directly limits agreement.
  • domain assumption The trained NEP models remain accurate for the long, large-scale MD trajectories at the studied temperatures and pressures.
    Ensemble RMSEs are reported (Section II A), but transferability to configurations visited in nanosecond-scale simulations is assumed.
  • domain assumption Classical MD statistics supplemented by the first-order quantum correction factor in Eq. (6) adequately approximate quantum vibrational statistics for INS intensities.
    Stated in Section II C; the paper shows the overtone at 250 meV and the combination mode at 550 meV in Sc-doped BaTiO3 are not captured, confirming the limitation.
  • domain assumption The one-dimensional Gaussian instrument resolution functions and kinematic masks are sufficient for quantitative predictions.
    The authors explicitly note that true resolution functions are four-dimensional and non-Gaussian but are used routinely (Section II C).
  • domain assumption PIMD with 32 beads and the stated simulation timescales provide converged spectra for benzene and Sc-doped BaTiO3.
    Used without a detailed convergence analysis in the main text (Section II B); the authors refer to supplementary sections for bead number choices.
  • domain assumption For Sc-doped BaTiO3, cubic and hexagonal structures are metastable on MD timescales, allowing composition and phase to be sampled independently.
    This is stated in Section III C and is needed to compare simulated spectra for both phases against experimental data from a single stable phase.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Predicting neutron experiments from first principles: A workflow powered by machine learning." pith.science (2026). https://pith.science/paper/ONS32TTP

@misc{pith2026250419352,
  author       = {Pith},
  title        = {Pith review of: Predicting neutron experiments from first principles: A workflow powered by machine learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ONS32TTP}},
  note         = {Machine review of arXiv:2504.19352}
}
read the original abstract

Machine learning has emerged as a powerful tool in materials discovery, enabling the rapid design of novel materials with tailored properties for countless applications, including in the context of energy and sustainability. To ensure the reliability of these methods, however, rigorous validation against experimental data is essential. Scattering techniques -- using neutrons, X-rays, or electrons -- offer a direct way to probe atomic-scale structure and dynamics, making them ideal for this purpose. In this work, we describe a computational workflow that bridges machine learning-based simulations with experimental validation. The workflow combines density functional theory, machine-learned interatomic potentials, molecular dynamics, and autocorrelation function analysis to simulate experimental signatures, with a focus on inelastic neutron scattering. We demonstrate the approach on three representative systems: crystalline silicon, crystalline benzene, and hydrogenated scandium-doped BaTiO3, comparing the simulated spectra to measurements from four different neutron spectrometers. While our primary focus is inelastic neutron scattering, the workflow is readily extendable to other modalities, including diffraction and quasi-elastic scattering of neutrons, X-rays, and electrons. The good agreement between simulated and experimental results highlights the potential of this approach for guiding and interpreting experiments, while also pointing out areas for further improvement.

Figures

Figures reproduced from arXiv: 2504.19352 by the authors.

Figure 1
Figure 1. FIG. 1. Workflow for simulating neutron scattering experiments from first principles. (a) The first step of the workflow [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. (a) Simulated INS dispersion of Si from MD for the ARCS spectrometer at spallation neutron source, with the [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Kinematic constraints for the four neutron instru [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: FIG. 4. (a) Simulated intensity at the X-point at 1500 K in [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. (a) Simulated INS spectra for crystalline benzene, at increasing levels of refinement, compared to an experimental [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Hydrogen dynamics in hydrogenated Sc-doped BaTiO [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. Contribution to the total dynamic structure factor [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. NEP89: Universal neuroevolution potential for inorganic and organic materials across 89 elements

    cond-mat.mtrl-sci 2025-04 conditional novelty 6.0 of 10

    NEP89 is a universal neuroevolution potential that combines near-DFT accuracy with empirical-potential speed for atomistic simulations of inorganic and organic materials across 89 elements.

Reference graph

Works this paper leans on

62 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    Shahzad, A

    K. Shahzad, A. I. Mardare, and A. W. Hassel, Ac- celerating materials discovery: Combinatorial synthe- sis, high-throughput characterization, and computational advances, Science and Technology of Advanced Materi- als: Methods 4, 2292486 (2024)

    work page 2024

  2. [2]

    C. Li, L. Bao, Y. Ji, Z. Tian, M. Cui, Y. Shi, Z. Zhao, and X. Wang, Combining machine learning and metal– organic frameworks research: Novel modeling, perfor- mance prediction, and materials discovery, Coordination Chemistry Reviews 514, 215888 (2024)

    work page 2024

  3. [3]

    C. Chen, D. T. Nguyen, S. J. Lee, N. A. Baker, A. S. Karakoti, L. Lauw, C. Owen, K. T. Mueller, B. A. Bilodeau, V. Murugesan, and M. Troyer, Accelerating Computational Materials Discovery with Machine Learn- ing and Cloud High-Performance Computing: From Large-Scale Screening to Experimental Validation, Jour- nal of the American Chemical Society 146, 20009 (2024)

    work page 2024

  4. [4]

    Fujii, J

    S. Fujii, J. Hyodo, K. Shitara, A. Kuwabara, S. Kasamatsu, and Y. Yamazaki, Emerging computa- tional and machine learning methodologies for proton- conducting oxides: Materials discovery and fundamental understanding, Science and Technology of Advanced Ma- terials 25, 2416383 (2024)

    work page 2024

  5. [5]

    Hautier, C

    G. Hautier, C. C. Fischer, A. Jain, T. Mueller, and G. Ceder, Finding Nature’s Missing Ternary Oxide Com- pounds Using Machine Learning and Density Functional Theory, Chemistry of Materials 22, 3762 (2010)

    work page 2010

  6. [6]

    Huang, F

    G. Huang, F. Huang, and W. Dong, Machine learning in energy storage material discovery and performance prediction, Chemical Engineering Journal 492, 152294 (2024)

    work page 2024

  7. [7]

    K. Guo, Z. Yang, C.-H. Yu, and M. J. Buehler, Artificial intelligence and machine learning in design of mechanical materials, Materials Horizons 8, 1153 (2021)

    work page 2021

  8. [8]

    Lookman, P

    T. Lookman, P. V. Balachandran, D. Xue, and R. Yuan, Active learning in materials science with emphasis on adaptive sampling using uncertainties for targeted de- sign, npj Computational Materials 5, 1 (2019)

    work page 2019

Show all 62 references
  1. [9]

    S. Wu, Y. Kondo, M.-a. Kakimoto, B. Yang, H. Yamada, I. Kuwajima, G. Lambard, K. Hongo, Y. Xu, J. Sh- iomi, C. Schick, J. Morikawa, and R. Yoshida, Machine- learning-assisted discovery of polymers with high ther- mal conductivity using a molecular design algorithm, npj Computa...

    2019

  2. [10]

    M. H. M¨ user, S. V. Sukhomlinov, and L. Pastewka, In- teratomic potentials: Achievements and challenges, Ad- vances in Physics: X 8, 2093129 (2023)

    2023

  3. [11]

    O. T. Unke, S. Chmiela, H. E. Sauceda, M. Gastegger, I. Poltavsky, K. T. Sch¨ utt, A. Tkatchenko, and K.-R. M¨ uller, Machine Learning Force Fields, Chemical Re- views 121, 10142 (2021)

    2021

  4. [12]

    Behler, Four Generations of High-Dimensional Neu- ral Network Potentials, Chemical Reviews 121, 10037 (2021)

    J. Behler, Four Generations of High-Dimensional Neu- ral Network Potentials, Chemical Reviews 121, 10037 (2021)

    2021

  5. [13]

    Z. Chen, N. Andrejevic, N. C. Drucker, T. Nguyen, R. P. Xian, T. Smidt, Y. Wang, R. Ernstorfer, D. A. Tennant, M. Chan, and M. Li, Machine learning on neutron and x- ray scattering and spectroscopies, Chemical Physics Re- views 2, 031301 (2021)

    2021

  6. [14]

    Ehlers, M

    G. Ehlers, M. L. Crow, Y. Diawara, F. X. Gallmeier, X. Geng, G. E. Granroth, R. D. Gregory, F. F. Islam, R. O. Knudson, F. Li, M. S. Loyd, and B. Vacaliuc, Mod- ern Trends in Neutron Scattering Instrument Technolo- gies, Instruments 6, 22 (2022)

    2022

  7. [15]

    Borgschulte, J

    A. Borgschulte, J. Terreni, E. Billeter, L. Daemen, Y. Cheng, A. Pandey, Z. Lodziana, R. J. Hemley, and A. J. Ramirez-Cuesta, Inelastic neutron scattering evi- dence for anomalous H–H distances in metal hydrides, Proceedings of the National Academy of Sciences 117, 4021 (2020)

    2020

  8. [16]

    Y. Q. Cheng, A. I. Kolesnikov, and A. J. Ramirez-Cuesta, Simulation of Inelastic Neutron Scattering Spectra Di- rectly from Molecular Dynamics Trajectories, Journal of Chemical Theory and Computation 16, 7702 (2020)

    2020

  9. [17]

    Z. Fan, Y. Wang, P. Ying, K. Song, J. Wang, Y. Wang, Z. Zeng, K. Xu, E. Lindgren, J. M. Rahm, A. J. 11 Gabourie, J. Liu, H. Dong, J. Wu, Y. Chen, Z. Zhong, J. Sun, P. Erhart, Y. Su, and T. Ala-Nissila, GPUMD: A package for constructing accurate machine-learned po- tentials and...

    2022

  10. [18]

    Fransson, M

    E. Fransson, M. Slabanja, P. Erhart, and G. Wahnstr¨ om, Dynasor—A Tool for Extracting Dynamical Structure Factors and Current Correlation Functions from Molec- ular Dynamics Simulations, Advanced Theory and Sim- ulations 4, 2000240 (2021)

    2021

  11. [19]

    Berger, E

    E. Berger, E. Fransson, F. Eriksson, E. Lindgren, G. Wahnstr¨ om, T. H. Rod, and P. Erhart, Dynasor 2: From Simulation to Experiment Through Correlation Functions (2025), arXiv:2503.21957 [cond-mat]

    2025 arXiv

  12. [20]

    Z. Fan, Z. Zeng, C. Zhang, Y. Wang, K. Song, H. Dong, Y. Chen, and T. Ala-Nissila, Neuroevolution machine learning potentials: Combining high accuracy and low cost in atomistic simulations and application to heat transport, Physical Review B 104, 104309 (2021)

    2021

  13. [21]

    Fan, Improving the accuracy of the neuroevolution machine learning potential for multi-component systems, Journal of Physics: Condensed Matter 34, 125902 (2022)

    Z. Fan, Improving the accuracy of the neuroevolution machine learning potential for multi-component systems, Journal of Physics: Condensed Matter 34, 125902 (2022)

    2022

  14. [22]

    Fransson, J

    E. Fransson, J. Wiktor, and P. Erhart, Phase Transitions in Inorganic Halide Perovskites from Machine-Learned Potentials, The Journal of Physical Chemistry C 127, 13773 (2023)

    2023

  15. [23]

    Lindgren, M

    E. Lindgren, M. Rahm, E. Fransson, F. Eriksson, N. ¨Osterbacka, Z. Fan, and P. Erhart, Calorine: A Python package for constructing and sampling neuroevo- lution potential models, Journal of Open Source Software 9, 6264 (2024)

    2024

  16. [24]

    A. H. Larsen, J. J. Mortensen, J. Blomqvist, I. E. Castelli, R. Christensen, M. Du lak, J. Friis, M. N. Groves, B. Hammer, C. Hargus, E. D. Hermes, P. C. Jennings, P. B. Jensen, J. Kermode, J. R. Kitchin, E. L. Kolsbjerg, J. Kubal, K. Kaasbjerg, S. Lysgaard, J. B. Maronsson, T...

    2017

  17. [25]

    Eriksson, E

    F. Eriksson, E. Fransson, and P. Erhart, The Hiphive Package for the Extraction of High-Order Force Con- stants by Machine Learning, Advanced Theory and Sim- ulations 2, 1800184 (2019)

    2019

  18. [26]

    Parrinello and A

    M. Parrinello and A. Rahman, Study of an F center in molten KCl, The Journal of Chemical Physics 80, 860 (1984)

    1984

  19. [27]

    P. Ying, W. Zhou, L. Svensson, E. Berger, E. Frans- son, F. Eriksson, K. Xu, T. Liang, J. Xu, B. Song, S. Chen, P. Erhart, and Z. Fan, Highly efficient path- integral molecular dynamics simulations with GPUMD using neuroevolution potentials: Case studies on thermal properties ...

    2025

  20. [28]

    Rossi, M

    M. Rossi, M. Ceriotti, and D. E. Manolopoulos, How to remove the spurious resonances from ring polymer molec- ular dynamics, The Journal of Chemical Physics 140, 234116 (2014)

    2014

  21. [29]

    R. Fair, A. Jackson, D. Voneshen, D. Jochym, D. Le, K. Refson, and T. Perring, Euphonic: Inelastic neutron scattering simulations from force constants and visual- ization tools for phonon properties, Journal of Applied Crystallography 55, 1689 (2022)

    2022

  22. [30]

    Turanyi, A

    R. Turanyi, A. Jackson, and J. Wilkins, Pace- neutrons/resins: Python library for resolution functions of inelastic neutron scattering instruments (2025), (ac- cessed 2025-04-10)

    2025

  23. [31]

    Dymkowski, S

    K. Dymkowski, S. F. Parker, F. Fernandez-Alonso, and S. Mukhopadhyay, AbINS: The modern software for INS interpretation, Physica B: Condensed Matter The 11th International Conference on Neutron Scattering (ICNS 2017), 551, 443 (2018)

    2018

  24. [32]

    Arnold, J

    O. Arnold, J. C. Bilheux, J. M. Borreguero, A. Buts, S. I. Campbell, L. Chapon, M. Doucet, N. Draper, R. F. Leal, M. A. Gigg, V. E. Lynch, A. Markvardsen, D. J. Mikkel- son, R. L. Mikkelson, R. Miller, K. Palmen, P. Parker, G. Passos, T. G. Perring, P. F. Peterson, S. Ren, M. ...

    2014

  25. [33]

    Rosander, E

    P. Rosander, E. Fransson, N. ¨Osterbacka, P. Erhart, and G. Wahnstr¨ om, Untangling the Raman spectra of cubic and tetragonal bazro3, Physical Review B 111, 064107 (2025)

    2025

  26. [34]

    Cardona and G

    M. Cardona and G. G¨ untherodt, eds.,Light Scattering in Solids II , Topics in Applied Physics, Vol. 50 (Springer, Berlin, Heidelberg, 1982)

    1982

  27. [35]

    P. E. Bl¨ ochl, Projector augmented-wave method, Physi- cal Review B 50, 17953 (1994)

    1994

  28. [36]

    Kresse and D

    G. Kresse and D. Joubert, From ultrasoft pseudopoten- tials to the projector augmented-wave method, Physical Review B 59, 1758 (1999)

    1999

  29. [37]

    Kresse and J

    G. Kresse and J. Hafner, Ab initio molecular dynamics for liquid metals, Physical Review B 47, 558 (1993)

    1993

  30. [38]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Computational Materials Science 6, 15 (1996)

    1996

  31. [39]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Physical Review B 54, 11169 (1996)

    1996

  32. [40]

    Berland and P

    K. Berland and P. Hyldgaard, Exchange functional that tests the robustness of the plasmon description of the van der Waals density functional, Physical Review B 89, 035412 (2014)

    2014

  33. [41]

    J. W. Furness, A. D. Kaplan, J. Ning, J. P. Perdew, and J. Sun, Accurate and Numerically Efficient r2SCAN Meta-Generalized Gradient Approximation, The Journal of Physical Chemistry Letters 11, 8208 (2020)

    2020

  34. [42]

    S. F. Parker, F. Fernandez-Alonso, A. J. Ramirez-Cuesta, J. Tomkinson, S. Rudic, R. S. Pinna, G. Gorini, and J. Fern´ andez Casta˜ non, Recent and future developments on TOSCA at ISIS, Journal of Physics: Conference Series 554, 012003 (2014)

    2014

  35. [43]

    R. S. Pinna, S. Rudi´ c, S. F. Parker, J. Armstrong, M. Zanetti, G. ˇSkoro, S. P. Waller, D. Zacek, C. A. Smith, M. J. Capstick, D. J. McPhail, D. E. Pooley, G. D. How- ells, G. Gorini, and F. Fernandez-Alonso, The neutron guide upgrade of the TOSCA spectrometer, Nuclear In- 1...

    2018

  36. [44]

    C. R. Harris, K. J. Millman, S. J. van der Walt, R. Gom- mers, P. Virtanen, D. Cournapeau, E. Wieser, J. Tay- lor, S. Berg, N. J. Smith, R. Kern, M. Picus, S. Hoyer, M. H. van Kerkwijk, M. Brett, A. Haldane, J. F. del R´ ıo, M. Wiebe, P. Peterson, P. G´ erard-Marchant, K. Shep...

    2020

  37. [45]

    Wes McKinney, Data Structures for Statistical Comput- ing in Python, inProceedings of the 9th Python in Science Conference, edited by St´ efan van der Walt and Jarrod Millman (2010) pp. 56 – 61

    2010

  38. [46]

    The pandas development team, Pandas-dev/pandas: Pandas, Zenodo (2024)

    2024

  39. [47]

    Plots were generated using matplotlib [48], with color maps from perfect-cmaps [49]

    packages. Plots were generated using matplotlib [48], with color maps from perfect-cmaps [49]. Atomic structures were visualized and analyzed using Ovito [50]. III. RESULTS A. Anharmonicity in Si We begin by applying the workflow outlined in the methodology section to simulate...

  40. [48]

    Virtanen, R

    P. Virtanen, R. Gommers, T. E. Oliphant, M. Haber- land, T. Reddy, D. Cournapeau, E. Burovski, P. Pe- terson, W. Weckesser, J. Bright, S. J. van der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, ˙I. Po- lat, Y...

    2020

  41. [49]

    J. D. Hunter, Matplotlib: A 2D Graphics Environment, Computing in Science & Engineering 9, 90 (2007)

    2007

  42. [50]

    Ulmestrand, Perfect-cmaps (2025), (accessed 2025-04- 10)

    M. Ulmestrand, Perfect-cmaps (2025), (accessed 2025-04- 10)

    2025

  43. [51]

    A. Stukowski, Visualization and analysis of atomistic simulation data with OVITO–the Open Visualization Tool, Modelling and Simulation in Materials Science and Engineering 18, 015012 (2009)

    2009

  44. [52]

    D. S. Kim, O. Hellman, J. Herriman, H. L. Smith, J. Y. Y. Lin, N. Shulumba, J. L. Niedziela, C. W. Li, D. L. Abernathy, and B. Fultz, Nuclear quantum effect with pure anharmonicity and the anomalous thermal ex- pansion of silicon, Proceedings of the National Academy of Science...

    2018

  45. [53]

    G. L. Squires, Introduction to the Theory of Thermal Neutron Scattering, 3rd ed. (Cambridge University Press, Cambridge, 2012)

    2012

  46. [54]

    Hellman and I

    O. Hellman and I. A. Abrikosov, Temperature-dependent effective third-order interatomic force constants from first principles, Physical Review B 88, 144301 (2013)

    2013

  47. [55]

    Tadano, Y

    T. Tadano, Y. Gohda, and S. Tsuneyuki, Anharmonic force constants extracted from first-principles molecu- lar dynamics: Applications to heat transfer simulations, Journal of Physics: Condensed Matter 26, 225402 (2014)

    2014

  48. [56]

    Monacelli, R

    L. Monacelli, R. Bianco, M. Cherubini, M. Calandra, I. Errea, and F. Mauri, The stochastic self-consistent har- monic approximation: Calculating vibrational properties of materials with full quantum and anharmonic effects, Journal of Physics: Condensed Matter 33, 363001 (2021)

    2021

  49. [57]

    Perrichon, N

    A. Perrichon, N. Torino, E. Jedvik Granhed, Y.-C. Lin, S. F. Parker, M. Jim´ enez-Ruiz, M. Karlsson, and P. F. Henry, Local Coordination Environments and Vibra- tional Dynamics of Protons in Hexagonal and Cubic Sc- Doped BaTiO3 Proton-Conducting Oxides, The Journal of Physical...

    2020

  50. [58]

    Chen and S

    C. Chen and S. P. Ong, A universal graph deep learn- ing interatomic potential for the periodic table, Nature Computational Science 2, 718 (2022)

    2022

  51. [59]

    B. Deng, P. Zhong, K. Jun, J. Riebesell, K. Han, C. J. Bartel, and G. Ceder, CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling, Nature Machine Intelligence 5, 1031 (2023)

    2023

  52. [60]

    Merchant, S

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

    2023

  53. [61]

    F. Xie, T. Lu, S. Meng, and M. Liu, GPTFF: A high- accuracy out-of-the-box universal AI force field for ar- bitrary inorganic materials, Science Bulletin 69, 3525 (2024)

    2024

  54. [62]

    Batatia, P

    I. Batatia, P. Benner, Y. Chiang, A. M. Elena, D. P. Kov´ acs, J. Riebesell, X. R. Advincula, M. Asta, M. Avaylon, W. J. Baldwin, F. Berger, N. Bernstein, A. Bhowmik, S. M. Blau, V. C˘ arare, J. P. Darby, S. De, F. D. Pia, V. L. Deringer, R. Elijoˇ sius, Z. El-Machachi, F. Fal...

    2024 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.