Pith. sign in

REVIEW 4 major objections 5 minor 59 references

Efficient and Accurate Machine Learning Interatomic Potential for Graphene: Capturing Stress-Strain and Vibrational Properties

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A machine-learned potential trained on tiny ab initio cells reproduces fracture, elasticity, and phonons in large graphene sheets.

desk verdict A useful, openly released DeepMD graphene potential, but the headline 'ab initio-level precision' for fracture outstrips the evidence since the tearing and stress–strain behavior are trained-in rather than independently predicted. read the letter →

arxiv 2505.12140 v1 pith:ROTMBU4F submitted 2025-05-17 cond-mat.mtrl-sci physics.comp-ph

Felipe Hawthorne , Paulo R. E. Raulino , Ronaldo Rodrigues Pelá , Cristiano F. Woellner This is my paper · ORCID
classification cond-mat.mtrl-sciphysics.comp-ph
keywords machinelearninginteratomicpotentialgraphenemoleculardynamicsstress-strainelasticconstantsphonondispersionvibrationaldensityofstatesfracture
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 claims that a machine-learning interatomic potential (MLIP) trained exclusively on ab initio molecular dynamics of tiny graphene cells (4 to 32 atoms) can replace density functional theory in large-scale molecular dynamics of graphene. The potential reproduces the stress-strain response up to fracture, the elastic constants, the phonon dispersion including the quadratic ZA mode, and the vibrational density of states, and it does so at computational cost that grows linearly with the number of atoms. If true, this gives a drop-in surrogate for DFT in simulations of roughly 9,000-atom graphene sheets, enabling studies of fracture, strain engineering, and phonon-limited transport that would otherwise be too expensive. The paper also claims the trained potential is reactive: it captures temperature-dependent tearing and the appearance of linear acetylenic carbon chains after fracture.

What carries the argument

The load-bearing object is a deep neural-network interatomic potential whose total energy is a sum of per-atom energies, each computed from a local environment descriptor. For atom $i$, neighbors inside a cutoff radius $r_c = 8$ Å are weighted by a switching function $s_{ij}$ that smoothly goes to zero at the cutoff, forming a coordinate matrix $R_i$; an embedding network turns $R_i$ into descriptor $D_i = G_i^T R_i R_i^T G_i$, and a second network maps it to $E_i$. Forces and the virial tensor are obtained by differentiating the total energy with respect to atomic positions and cell deformations, so the model can be trained against energy, force, and stress labels simultaneously. This construction is what lets a potential trained on small strained cells carry over to slow fracture and thermal vibrations in large sheets.

What would settle it

Retrain the exact same architecture on the same AIMD dataset with every NPT frame that shows torn edges or carbon-chain fragments removed, then deform a pristine 9,072-atom sheet at 300 K. If the linear acetylenic chains no longer appear, the paper's claim that they emerge during tearing is actually a memory of the training data; if they still appear, the potential is generating new reactive configurations beyond its training set.

Watch

Extended reading notes

Core claim

The paper's central discovery is that a deep neural-network potential with local environment descriptors can serve as a reactive surrogate for density functional theory across both mechanical and vibrational response of graphene. Trained on AIMD frames from 4-, 8-, 16-, and 32-atom cells in NVT, NPT, and NVE ensembles with strain rates from $10^{-6}$ to $10^{-2}$ fs$^{-1}$, the model reproduces, in a 9,072-atom sheet, the stress-strain curves at 1 K and 300 K; the elastic constants ($C_{11} = 1.093$ TPa, $C_{21} = 0.199$ TPa, shear modulus $0.446$ TPa, Poisson's ratio $0.183$, Young's modulus $1.056$ TPa); the phonon dispersion with a quadratic ZA branch; and the vibrational density of states with strain-induced red shifts in-plane and a blue shift out-of-plane. It also reports the appearance of linear acetylenic carbon chains in the torn regions after fracture, at all studied temperatures and more prominently under $x$-direction strain.

Load-bearing premise

The model's transferability stands or falls on the premise that a few short, sub-5-picosecond LDA trajectories of 4- to 32-atom cells, many of which are strained until they tear, cover the same potential energy surface that a 9,072-atom sheet explores during slow fracture and vibration.

Editorial extensions

If this is right

  • A single potential can now replace separate classical force fields for strained and vibrating graphene, so simulations that need both mechanical and thermal response no longer have to switch models.
  • Fracture studies on graphene sheets of thousands of atoms become feasible at DFT-level fidelity, including crack-tip chemistry and the formation of carbon-chain products.
  • The strain-dependent shifts in the vibrational spectrum predicted by the model line up with Raman experiments, so the potential can guide strain-engineering experiments before they are run.
  • Because the potential reproduces both 1 K and 300 K failure, temperature-dependent strength and ductility can be studied in large sheets without expensive ab initio molecular dynamics.
  • The described training workflow, with data and parameters made public, gives other groups a template for building similar reactive potentials for new two-dimensional materials.

Reading between the lines

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

  • Inference: because the training data are LDA-based, the model's quantitative elastic constants plausibly inherit LDA's well-known tendency to overbind; retraining on a different exchange-correlation functional would show how much of the DFT agreement is functional-specific.
  • Inference: the linear acetylenic chains seen in the large sheet may be replaying torn configurations already present in the small-cell NPT training runs; deleting those frames and retraining is a clean test of whether the chains are emergent.
  • Inference: the same descriptor-plus-loss recipe is portable to other two-dimensional materials, but portability is not guaranteed by architecture alone; the training set must include the same breadth of strain rates, ensembles, and near-fracture geometries.
  • Inference: the VDOS is computed from only 1 ps of velocity autocorrelation, so low-frequency features near the ZA mode are likely under-resolved; longer correlation windows would sharpen them and test the model in precisely the regime its training data sample least.
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

4 major / 5 minor

Summary. This manuscript presents a DeepMD machine-learning interatomic potential for graphene trained on LDA AIMD trajectories of 4-, 8-, 16-, and 32-atom cells under NVT, NPT, and NVE conditions, including strained and torn configurations. The authors report RMSE validation of energies, forces, and virial stress; linear-scaling LAMMPS performance; stress-strain curves for a 9072-atom sheet at 1 K and 300 K that show temperature- and direction-dependent fracture; the appearance of linear acetylenic carbon (LAC) chains upon tearing; elastic constants and derived Young's modulus and Poisson's ratio; phonon dispersion; and strain-dependent vibrational density of states. The stated claims are that the potential reproduces mechanical and vibrational properties with ab initio-level precision and can serve as a transferable surrogate for DFT in large-scale graphene molecular dynamics.

Significance. If fully supported, the potential would be a useful open resource for graphene MD: the elastic constants are in reasonable agreement with several DFT and experimental references, the phonon dispersion has the expected qualitative structure, and the public release of training data and model parameters on NOMAD and GitHub is a clear strength that supports reproducibility. The linear-scaling benchmark is also a useful practical result. However, the load-bearing evidence for the fracture and transferability claims is incomplete: the training set already contains torn high-strain configurations and the loss function includes virial stress, so the 'emergent' LAC-chain and fracture behavior is at least partly fitted rather than independently predicted; the stress-strain validation is qualitative; and the vibrational validation is largely visual. The contribution is real, but the claimed scope exceeds what the evidence currently establishes.

major comments (4)
  1. [Methods, Section 2; Supporting Information Table S1] The training dataset already includes torn graphene structures: the authors state 'During these latter runs, tearing of the graphene structures was observed' for NPT runs with strain rates up to 10^-2 fs^-1, and the 32-atom entries in Table S1 include 5x10^-3 and 10^-2 fs^-1 runs. Since the loss function also fits the virial stress, the fracture behavior and LAC chains in Figures 4 and 5 are at least in part direct reproduction of training configurations, not emergent predictions. The abstract's wording 'emergence of linear acetylenic carbon chains upon tearing' and the conclusion's characterization of LAC formation as 'an emergent phenomenon' are therefore not supported unless a test is provided, e.g., retraining without tearing frames or a quantitative comparison of the large-sheet fracture path with AIMD data.
  2. [Section 3, Figure 4] The stress-strain curves are discussed only qualitatively in terms of temperature and direction trends relative to Ref. 42. No quantitative comparison of fracture stress, fracture strain, or post-yield stress-strain path with DFT or experiment is provided, despite the abstract claiming 'ab initio-level precision' for mechanical deformation. Please add quantitative benchmarks for the 9072-atom sheet, including ideal strength and fracture strain at the simulated strain rates, with appropriate error estimates and a discussion of strain-rate effects.
  3. [Section 3, Table 2] The elastic validation covers only the near-equilibrium linear regime, which is interpolation from the small-strain AIMD training data. The reported C11 and C21 values differ from the cited MD/MM results by large margins (e.g., C21 differs by roughly 20-46% from the AMBER, Tersoff, and MM values), while only Young's modulus and Poisson's ratio agree well with experiment and DFT. This table does not validate the fracture behavior highlighted in the abstract, and the phrase 'ab initio-level precision' should be restricted to the properties actually benchmarked unless additional anharmonic/fracture validation is provided.
  4. [Section 3, Figures 6 and 7] The 'excellent agreement' with experimental phonon data is based on visual comparison, and the quadratic ZA mode near the Gamma point is a symmetry-dictated feature of a 2D membrane rather than a stringent test of the potential. Similarly, the VDOS peak shifts are compared qualitatively to Raman literature; the reported in-plane red shift of about 100 cm^-1 at 3% strain appears substantially larger than typical experimental G-band shifts for uniaxial strain, so a quantitative comparison with experimental or DFT-computed strain coefficients is needed. Please report quantitative phonon-frequency errors at high-symmetry points and a numerical comparison of strain-dependent VDOS peak positions.
minor comments (5)
  1. [Introduction] There is a typo in the first paragraph: 'sytems' should be 'systems'.
  2. [Methods, MLIP training paragraph] The phrase 'a total of 108 steps' should likely read '10^8 steps'; please use standard mathematical formatting for the learning schedule.
  3. [Methods, Eq. (2)] The switching function definition is confusing because the first branch also contains 1/r_ij and the variable r is not defined relative to r_ij; please clarify the notation and ensure the switching function is continuous at r_s and r_c.
  4. [Figure 5] Figure 5 colors atoms by von Mises stress but no color bar or scale is provided, making the snapshots difficult to interpret quantitatively.
  5. [Figure 4] The stress-strain figure caption does not state the units or the axis labels; please include them explicitly and note which curve corresponds to which orientation.

Circularity Check

1 steps flagged · score 6.0 of 10

Fracture and LAC-chain 'emergence' is fitted from training trajectories that already contain tearing; phonon and VDOS results retain independent content.

  1. fitted input called prediction [Section 2 (Methods, AIMD); Section 3 (Results, Strain Engineering); Section 4 (Conclusions)]
    "During these latter runs, tearing of the graphene structures was observed. ... the formation of linear acetylenic carbon (LAC) chains during tearing—an emergent phenomenon not trivially reproduced by traditional force fields."

    The AIMD training set explicitly includes NPT trajectories in which tearing occurs, and the DeepMD loss fits forces and the virial tensor on those torn frames. The large-sheet fracture curves and LAC chains are therefore generated by a potential fitted to small-cell torn configurations; the advertised 'emergence' of LAC chains is a reproduction of training-regime behavior, not an independent discovery. The stress–strain response is likewise tied to the fitted virial, though external elastic-constant benchmarks provide some independent content.

full rationale

The core workflow—fit DeepMD to AIMD energies, forces, and virials, then run MD—is not inherently circular; the paper's RMSE table and comparisons to external DFT/experiment for elastic constants, phonon dispersion, and VDOS provide independent evidence that the fit generalizes. The one genuinely circular element is the presentation of fracture behavior and LAC chain formation as an 'emergent phenomenon': the Methods state that the AIMD NPT training runs already exhibited tearing, and the loss directly fits forces and virial on those configurations. Thus the large-sheet stress–strain curves and torn-chain morphologies are interpolations of the training set, and calling them emergent overstates the novelty. This is partial circularity, not total: the phonon dispersion and strain-induced VDOS shifts are trajectory/spectral quantities not entered as direct labels, and elastic constants are checked against external benchmarks. No load-bearing self-citation chain exists; self-citations 58–59 are data/code availability only.

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

The paper's claims rest on the training data quality, the DeepMD model capacity, and the equivalence between model observables and experimental quantities. No new physical entities are introduced. The trained network weights are many fitted parameters, but they are fitted to AIMD reference data, not to the target properties; the hand-chosen hyperparameters are the relevant free choices.

free parameters (5)
  • Cutoff radius rc = 8 Å
    Chosen by hand (Methods, Section 2). It defines the local atomic environment; a larger cutoff would change descriptors and accuracy.
  • Descriptor size and embedding network widths = 16; 50, 100, 200 neurons
    Architecture hyperparameters (Methods). These control the expressiveness of the atomic descriptors.
  • Fitting network width and depth = 3 x 240 neurons
    Architecture hyperparameters (Methods) controlling the energy model capacity.
  • Loss pre-factor schedule = start (1e-2, 1, 1e3); final (1, 1e4, 10) for energy, force, virial
    Training schedule (Methods). It balances the errors on different targets and affects the final model.
  • Learning rate schedule = 1e-4 initial, decay 5e-6 per 5e4 steps
    Training hyperparameter (Methods) affecting convergence.
assumptions (4)
  • domain assumption LDA reliably describes graphene interatomic interactions for the strained, torn, and thermal configurations used in training.
    The training set is generated with GPAW using LDA (Methods). The paper justifies this by citing Ref 25 for reasonable agreement, but LDA accuracy for anharmonic and fracture configurations is not established.
  • domain assumption DeepMD with local two-body embedding descriptors can represent the graphene PES to the needed accuracy.
    The model capacity and descriptor are fixed before training (Methods); the paper does not benchmark alternative descriptors.
  • domain assumption AIMD trajectories of 4-32 atom cells over 0.5-5 ps with strain rates up to 1e-2 fs^-1 sample the phase space relevant to large-scale fracture and vibrational response.
    The transferability claim rests on this sampling sufficiency; no convergence tests with respect to cell size or simulation time are shown.
  • domain assumption The VDOS computed from velocity autocorrelation is comparable to experimental Raman spectra for the purpose of peak shift comparison.
    The paper compares VDOS peak shifts to experimental Raman G-band shifts (Section 3, Vibrational Analysis) without a projection onto Raman-active modes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient and Accurate Machine Learning Interatomic Potential for Graphene: Capturing Stress-Strain and Vibrational Properties." pith.science (2026). https://pith.science/paper/ROTMBU4F

@misc{pith2026250512140,
  author       = {Pith},
  title        = {Pith review of: Efficient and Accurate Machine Learning Interatomic Potential for Graphene: Capturing Stress-Strain and Vibrational Properties},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ROTMBU4F}},
  note         = {Machine review of arXiv:2505.12140}
}
read the original abstract

Machine learning interatomic potentials (MLIPs) offer an efficient and accurate framework for large-scale molecular dynamics (MD) simulations, effectively bridging the gap between classical force fields and \textit{ab initio} methods. In this work, we present a reactive MLIP for graphene, trained on an extensive dataset generated via \textit{ab initio} molecular dynamics (AIMD) simulations. The model accurately reproduces key mechanical and vibrational properties, including stress-strain behavior, elastic constants, phonon dispersion, and vibrational density of states. Notably, it captures temperature-dependent fracture mechanisms and the emergence of linear acetylenic carbon chains upon tearing. The phonon analysis also reveals the expected quadratic ZA mode and excellent agreement with experimental and DFT benchmarks. Our MLIP scales linearly with system size, enabling simulations of large graphene sheets with \textit{ab initio}-level precision. This work delivers a robust and transferable MLIP, alongside an accessible training workflow that can be extended to other materials.

Figures

Figures reproduced from arXiv: 2505.12140 by the authors.

Figure 1
Figure 1. (a) Illustration of the 32-C graphene structure used in the AIMD sim￾ulations. (b) Generation of the coordi￾nate matrix, R, and subsequent calcula￾tion of the descriptors, Di . The first el￾ement of Ri , sij , is illustrated. (c) These descriptors served as input elements for the NN, trained to predict the total energy E = P i Ei . (d) The trained MLIP was deployed as a force field in large-scale MD simulations. The… view at source ↗
Figure 3
Figure 3. Average time required for 100 NPT steps as a function of system size. The trained MLIP exhibits linear scaling behavior, O(N),as indicated by the dashed line. temperatures (1 and 300 K), with stress applied independently along both direc￾tions. The MLIP successfully reproduces two expected results related to tempera￾ture and strain direction.42 The tempera￾ture effect is evident in the difference in ultimate tensile… view at source ↗
Figure 2
Figure 2. Comparison between AIMD ref￾erence data and NN predictions (black scattered points), derived using AIMD simulation frames from the 32-C system as input: (a) energy per atom, (b) x￾component of atomic forces, and (c) xx￾component of the virial stress tensor. Red dashed lines indicate perfect agreement [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Stress-strain curves obtained us￾ing the MLIP. Solid lines correspond to simulations at T = 1 K, while dashed lines represent results at T = 300 K. mation of linear acetylenic carbon (LAC) chains43,44 after the graphene sheets frac￾tured [PITH_FULL_IMAGE:figures/full_…
Figure 5
Figure 5. Figure 5: Snapshots from strain simula￾tions for T = 300 K, with strain (a) ϵyy and (b) ϵxx. Atoms are colored according to their Von Mises stress. Even though LACs exhibit mechan￾ical properties that differ significantly from those of graphene,45 their formation has been closel…
Figure 6
Figure 6. Figure 6: Phonon dispersion of graphene obtained using the MLIP in LAMMPS. The in-plane (TO) and out-of-plane (ZO, ZA) phonon branches are labeled ac￾cordingly. Notably, the ZA vibrational mode exhibits quadratic behavior around Γ point. The scattered open circles repre￾sent exp…
Figure 7
Figure 7. Figure 7: Cv(t) (left) and corresponding VDOS (right) for vx (a)-(b), vy (c)-(d), and vz (e)-(f). The values in all panels are shifted along the y-axis for clarity. The color scheme indicates strain applied along ⃗y, as shown in panel (a). vious investigations on graphene’s ther…
Figure 1
Figure 1. Figure 1: Illustrations of the graphene structures used in the AIMD simulations, con [PITH_FULL_IMAGE:figures/full_fig_p017_1.png]
Figure 2
Figure 2. Figure 2: Corresponding velocity autocorrelation functions and VDOS (left and right [PITH_FULL_IMAGE:figures/full_fig_p019_2.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 55 canonical work pages

  1. [1]

    No free lunch theorems for optimization

    Wolpert, D.; Macready, W. No free lunch theorems for optimization. IEEE Transactions on Evolutionary Computation 1997, 1, 67--82

    work page 1997

  2. [2]

    T.; Frost, J

    Butler, K. T.; Frost, J. M.; Skelton, J. M.; Svane, K. L.; Walsh, A. Computational materials design of crystalline solids. Chemical Society Reviews 2016, 45, 6138--6146

    work page 2016

  3. [3]

    Extending Solid-State Calculations to Ultra-Long-Range Length Scales

    Müller, T.; Sharma, S.; Gross, E.; Dewhurst, J. Extending Solid-State Calculations to Ultra-Long-Range Length Scales. Physical Review Letters 2020, 125, 256402

    work page 2020

  4. [4]

    Pushing the limit of molecular dynamics with ab initio accuracy to 100 million atoms with machine learning

    Jia, W.; Wang, H.; Chen, M.; Lu, D.; Lin, L.; Car, R.; E, W.; Zhang, L. Pushing the limit of molecular dynamics with ab initio accuracy to 100 million atoms with machine learning. Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis. 2020

    work page 2020

  5. [5]

    Iftimie, R.; Minary, P.; Tuckerman, M. E. Ab initio molecular dynamics: Concepts, recent developments, and future trends. Proceedings of the National Academy of Sciences 2005, 102, 6654--6659

    work page 2005

  6. [6]

    P.; Hong, S.; Islam, M

    Senftle, T. P.; Hong, S.; Islam, M. M.; Kylasa, S. B.; Zheng, Y.; Shin, Y. K.; Junkermeier, C.; Engel-Herbert, R.; Janik, M. J.; Aktulga, H. M.; Verstraelen, T.; Grama, A.; van Duin, A. C. T. The ReaxFF reactive force-field: development, applications and future directions. npj Computational Materials 2016, 2

    work page 2016

  7. [7]

    K.; Casewit, C

    Rappe, A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A.; Skiff, W. M. UFF, a full periodic table force field for molecular mechanics and molecular dynamics simulations. Journal of the American Chemical Society 1992, 114, 10024--10035

    1992

  8. [8]

    V.; Woellner, C

    de Lima, J.; Bastos, L. V.; Woellner, C. F. A comparative study of the electronic and thermal properties of -graphyne and -graphdiyne nanoscrolls: Implications for thermoelectric applications. Computational Materials Science 2025, 250, 113728

    work page 2025

Show all 59 references
  1. [9]

    Characterizing mechanical properties of graphite using molecular dynamics simulation

    Tsai, J.-L.; Tu, J.-F. Characterizing mechanical properties of graphite using molecular dynamics simulation. Materials and Design 2010, 31, 194--199

    2010

  2. [10]

    Machine-learned potentials for next-generation matter simulations

    Friederich, P.; Häse, F.; Proppe, J.; Aspuru-Guzik, A. Machine-learned potentials for next-generation matter simulations. Nature Materials 2021, 20, 750--761

    2021

  3. [11]

    S.; Nebgen, B.; Messerly, R.; Li, Y

    Fedik, N.; Zubatyuk, R.; Kulichenko, M.; Lubbers, N.; Smith, J. S.; Nebgen, B.; Messerly, R.; Li, Y. W.; Boldyrev, A. I.; Barros, K.; Isayev, O.; Tretiak, S. Extending machine learning beyond interatomic potentials for predicting molecular properties. Nature Reviews Chemistry ...

    2022

  4. [12]

    Machine learning interatomic potential: Bridge the gap between small-scale models and realistic device-scale simulations

    Wang, G.; Wang, C.; Zhang, X.; Li, Z.; Zhou, J.; Sun, Z. Machine learning interatomic potential: Bridge the gap between small-scale models and realistic device-scale simulations. iScience 2024, 27, 109673

    2024

  5. [13]

    T.; Chmiela, S.; Sauceda, H

    Unke, O. T.; Chmiela, S.; Sauceda, H. E.; Gastegger, M.; Poltavsky, I.; Schütt, K. T.; Tkatchenko, A.; Müller, K.-R. Machine Learning Force Fields. Chemical Reviews 2021, 121, 10142--10186

    2021

  6. [14]

    R.; Lahiri, I.; Suresh, K

    Urade, A. R.; Lahiri, I.; Suresh, K. S. Graphene Properties, Synthesis and Applications: A Review. JOM 2022, 75, 614--630

    2022

  7. [15]

    Development of a machine learning potential for graphene

    Rowe, P.; Csányi, G.; Alfè, D.; Michaelides, A. Development of a machine learning potential for graphene. Physical Review B 2018, 97, 054303

    2018

  8. [16]

    Reliable machine learning potentials based on artificial neural network for graphene

    Singh, A.; Li, Y. Reliable machine learning potentials based on artificial neural network for graphene. Computational Materials Science 2023, 227, 112272

    2023

  9. [17]

    L.; Cai, C.; Lin, Y.; Wang, B.; Xu, J.; Zhu, J.-X.; Luo, C.; Zhang, Y.; Goodall, R

    Zeng, J.; Zhang, D.; Lu, D.; Mo, P.; Li, Z.; Chen, Y.; Rynik, M.; Huang, L.; Li, Z.; Shi, S.; Wang, Y.; Ye, H.; Tuo, P.; Yang, J.; Ding, Y.; Li, Y.; Tisi, D.; Zeng, Q.; Bao, H.; Xia, Y.; Huang, J.; Muraoka, K.; Wang, Y.; Chang, J.; Yuan, F.; Bore, S. L.; Cai, C.; Lin, Y.; Wang...

    2023

  10. [18]

    Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics

    Zhang, L.; Han, J.; Wang, H.; Car, R.; E, W. Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics. Physical Review Letters 2018, 120, 143001

    2018

  11. [19]

    P.; Aktulga, H

    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.; Shan, R.; Stevens, M. J.; Tranchida, J.; Trott, C.; Plimpton, S. J. LAMMPS - a flexible simulation tool for particle-b...

    2022

  12. [20]

    Blöchl, P. E. Projector augmented-wave method. Physical Review B 1994, 50, 17953--17979

    1994

  13. [21]

    J.; Chen, J.; Dułak, M.; Ferrighi, L.; Gavnholt, J.; Glinsvad, C.; Haikola, V.; Hansen, H

    Enkovaara, J.; Rostgaard, C.; Mortensen, J. J.; Chen, J.; Dułak, M.; Ferrighi, L.; Gavnholt, J.; Glinsvad, C.; Haikola, V.; Hansen, H. A.; Kristoffersen, H. H.; Kuisma, M.; Larsen, A. H.; Lehtovaara, L.; Ljungberg, M.; Lopez-Acevedo, O.; Moses, P. G.; Ojanen, J.; Olsen, T.; Pe...

    2010

  14. [22]

    afer, C.; J\'onsson, E. O.; Hermes, E. D.; Nilsson, F. A.; Kastlunger, G.; Levi, G.; J\'onsson, H.; H\

    Mortensen, J. J. r.; Larsen, A. H.; Kuisma, M.; Ivanov, A. V.; Taghizadeh, A.; Peterson, A.; Haldar, A.; Dohn, A. O.; Sch\"afer, C.; J\'onsson, E. O.; Hermes, E. D.; Nilsson, F. A.; Kastlunger, G.; Levi, G.; J\'onsson, H.; H\"akkinen, H.; Fojt, J.; Kangsabanik, J.; S dequist, ...

    2024

  15. [23]

    E.; Christensen, R.; Dułak, M.; Friis, J.; Groves, M

    Hjorth Larsen, A.; Jørgen Mortensen, J.; Blomqvist, J.; Castelli, I. E.; Christensen, R.; Dułak, M.; Friis, J.; Groves, M. N.; Hammer, B.; Hargus, C.; Hermes, E. D.; Jennings, P. C.; Bjerre Jensen, P.; Kermode, J.; Kitchin, J. R.; Leonhard Kolsbjerg, E.; Kubal, J.; Kaasbjerg, ...

    2017

  16. [24]

    Ziesche, P.; Kurth, S.; Perdew, J. P. Density functionals from LDA to GGA. Computational Materials Science 1998, 11, 122--127

    1998

  17. [25]

    Wang, Y.; Xu, Z.; Moe, Y. N. On the performance of local density approximation in describing the adsorption of electron donating/accepting molecules on graphene. Chemical Physics 2012, 406, 78--85

    2012

  18. [26]

    Understanding molecular simulation, 2nd ed.; Computational science series 1; Acad

    Frenkel, D.; Smit, B. Understanding molecular simulation, 2nd ed.; Computational science series 1; Acad. Press: San Diego [u.a.], 2011; Hier auch später erschienene, unveränderte Nachdrucke

    2011

  19. [27]

    Hoover NPT dynamics for systems varying in shape and size

    Melchionna, S.; Ciccotti, G.; Lee Holian, B. Hoover NPT dynamics for systems varying in shape and size. Molecular Physics 1993, 78, 533--544

    1993

  20. [28]

    Constrained systems and statistical distribution

    Melchionna, S. Constrained systems and statistical distribution. Physical Review E 2000, 61, 6165--6170

    2000

  21. [29]

    End-to-end Symmetry Preserving Inter-atomic Potential Energy Model for Finite and Extended Systems

    Zhang, L.; Han, J.; Wang, H.; Saidi, W.; Car, R.; E, W. End-to-end Symmetry Preserving Inter-atomic Potential Energy Model for Finite and Extended Systems. Advances in Neural Information Processing Systems. 2018

    2018

  22. [30]

    In Computer Vision - ECCV 2016; Matas, J., Sebe, N., Welling, M., Eds.; Lecture Notes in Computer Science Ser

    Leibe, B. In Computer Vision - ECCV 2016; Matas, J., Sebe, N., Welling, M., Eds.; Lecture Notes in Computer Science Ser. v.9908; Springer International Publishing AG: Cham, 2016; Description based on publisher supplied metadata and other sources

    2016

  23. [31]

    Deep Learning with Python, Second Edition; Manning Publications Co

    Chollet, F. Deep Learning with Python, Second Edition; Manning Publications Co. LLC: New York, 2021; Description based upon print version of record

    2021

  24. [32]

    Application of atomic stress to compute heat flux via molecular dynamics for systems with many-body interactions

    Surblys, D.; Matsubara, H.; Kikugawa, G.; Ohara, T. Application of atomic stress to compute heat flux via molecular dynamics for systems with many-body interactions. Physical Review E 2019, 99, 051301

    2019

  25. [33]

    Methodology and meaning of computing heat flux via atomic stress in systems with constraint dynamics

    Surblys, D.; Matsubara, H.; Kikugawa, G.; Ohara, T. Methodology and meaning of computing heat flux via atomic stress in systems with constraint dynamics. Journal of Applied Physics 2021, 130

    2021

  26. [34]

    Implementation strategies in phonopy and phono3py

    Togo, A.; Chaput, L.; Tadano, T.; Tanaka, I. Implementation strategies in phonopy and phono3py. J. Phys. Condens. Matter 2023, 35, 353001

    2023

  27. [35]

    M.; PASKIN, A

    DICKEY, J. M.; PASKIN, A. Computer Simulation of the Lattice Dynamics of Solids. Physical Review 1969, 188, 1407--1418

    1969

  28. [36]

    Computing vibrational spectra from ab initio molecular dynamics

    Thomas, M.; Brehm, M.; Fligg, R.; Vöhringer, P.; Kirchner, B. Computing vibrational spectra from ab initio molecular dynamics. Physical Chemistry Chemical Physics 2013, 15, 6608

    2013

  29. [37]

    Velocity-autocorrelation spectrum of simple classical liquids

    Bosse, J.; Götze, W.; Zippelius, A. Velocity-autocorrelation spectrum of simple classical liquids. Physical Review A 1978, 18, 1214--1221

    1978

  30. [38]

    H.; Yu, T.; Lu, Y

    Ni, Z. H.; Yu, T.; Lu, Y. H.; Wang, Y. Y.; Feng, Y. P.; Shen, Z. X. Uniaxial Strain on Graphene: Raman Spectroscopy Study and Band-Gap Opening. ACS Nano 2009, 3, 483--483

    2009

  31. [39]

    Investigation of strain-induced modulation on electronic properties of graphene field effect transistor

    Dong, J.; Liu, S.; Fu, Y.; Wang, Q. Investigation of strain-induced modulation on electronic properties of graphene field effect transistor. Physics Letters A 2017, 381, 292--297

    2017

  32. [40]

    Das, A.; Chakraborty, B.; Sood, A. K. Raman spectroscopy of graphene on different substrates and influence of defects. Bulletin of Materials Science 2008, 31, 579--584

    2008

  33. [41]

    R.; Millman, K

    Harris, C. R.; Millman, K. J.; van der Walt, S. J.; Gommers, R.; Virtanen, P.; Cournapeau, D.; Wieser, E.; Taylor, J.; Berg, S.; Smith, N. J.; Kern, R.; Picus, M.; Hoyer, S.; van Kerkwijk, M. H.; Brett, M.; Haldane, A.; del R \' i o, J. F.; Wiebe, M.; Peterson, P.; G \' e rard...

    2020

  34. [42]

    Zhao, H.; Aluru, N. R. Temperature and strain-rate dependent fracture strength of graphene. Journal of Applied Physics 2010, 108

    2010

  35. [43]

    Bryce, M. R. A review of functional linear carbon chains (oligoynes, polyynes, cumulenes) and their applications as molecular wires in molecular electronics and optoelectronics. Journal of Materials Chemistry C 2021, 9, 10524--10546

    2021

  36. [44]

    A review of linear carbon chains

    Zhang, K.; Zhang, Y.; Shi, L. A review of linear carbon chains. Chinese Chemical Letters 2020, 31, 1746--1756

    2020

  37. [45]

    Mechanical properties of carbyne: experiment and simulations

    Kotrechko, S.; Mikhailovskij, I.; Mazilova, T.; Sadanov, E.; Timoshevskii, A.; Stetsenko, N.; Matviychuk, Y. Mechanical properties of carbyne: experiment and simulations. Nanoscale Research Letters 2015, 10

    2015

  38. [46]

    Preparation of long monatomic carbon chains: Molecular dynamics studies

    Wang, Y.; Ning, X.-J.; Lin, Z.-Z.; Li, P.; Zhuang, J. Preparation of long monatomic carbon chains: Molecular dynamics studies. Physical Review B 2007, 76, 165423

    2007

  39. [47]

    Perpendicular growth of carbon chains on graphene from first-principles

    Ataca, C.; Ciraci, S. Perpendicular growth of carbon chains on graphene from first-principles. Physical Review B 2011, 83, 235417

    2011

  40. [48]

    D.; Rajendran, S.; Liew, K

    Reddy, C. D.; Rajendran, S.; Liew, K. M. Equilibrium configuration and continuum elastic properties of finite sized graphene. Nanotechnology 2006, 17, 864--870

    2006

  41. [49]

    Research on the auxetic behavior and mechanical properties of periodically rotating graphene nanostructures

    Liang, Y.; Huang, J.; Qu, J.; Huang, J.; Hui, D. Research on the auxetic behavior and mechanical properties of periodically rotating graphene nanostructures. Nanotechnology Reviews 2022, 11, 1733--1743

    2022

  42. [50]

    Numerical Investigation of the Fracture Mechanism of Defective Graphene Sheets

    Fan, N.; Ren, Z.; Jing, G.; Guo, J.; Peng, B.; Jiang, H. Numerical Investigation of the Fracture Mechanism of Defective Graphene Sheets. Materials 2017, 10, 164

    2017

  43. [51]

    Ab initio calculation of ideal strength and phonon instability of graphene under tension

    Liu, F.; Ming, P.; Li, J. Ab initio calculation of ideal strength and phonon instability of graphene under tension. Physical Review B 2007, 76, 064120

    2007

  44. [52]

    Raman Spectroscopy of nanomaterials: How spectra relate to disorder, particle size and mechanical properties

    Gouadec, G.; Colomban, P. Raman Spectroscopy of nanomaterials: How spectra relate to disorder, particle size and mechanical properties. Progress in Crystal Growth and Characterization of Materials 2007, 53, 1--56

    2007

  45. [53]

    NONLINEAR MECHANICS OF SINGLE-ATOMIC-LAYER GRAPHENE SHEETS

    LU, Q.; HUANG, R. NONLINEAR MECHANICS OF SINGLE-ATOMIC-LAYER GRAPHENE SHEETS. International Journal of Applied Mechanics 2009, 01, 443--467

    2009

  46. [54]

    Analysis of phonons in graphene sheets by means of HREELS measurement and ab initio calculation

    Yanagisawa, H.; Tanaka, T.; Ishida, Y.; Matsue, M.; Rokuta, E.; Otani, S.; Oshima, C. Analysis of phonons in graphene sheets by means of HREELS measurement and ab initio calculation. Surface and Interface Analysis 2005, 37, 133--136

    2005

  47. [55]

    Thermal Conduction Across Graphene Cross-Linkers

    Liu, X.; Zhang, G.; Zhang, Y.-W. Thermal Conduction Across Graphene Cross-Linkers. The Journal of Physical Chemistry C 2014, 118, 12541--12547

    2014

  48. [56]

    Raman spectroscopy of graphene-based materials and its applications in related devices

    Wu, J.-B.; Lin, M.-L.; Cong, X.; Liu, H.-N.; Tan, P.-H. Raman spectroscopy of graphene-based materials and its applications in related devices. Chemical Society Reviews 2018, 47, 1822--1873

    2018

  49. [57]

    The Raman redshift of graphene impacted by gold nanoparticles

    Zheng, X.; Chen, W.; Wang, G.; Yu, Y.; Qin, S.; Fang, J.; Wang, F.; Zhang, X.-A. The Raman redshift of graphene impacted by gold nanoparticles. AIP Advances 2015, 5

    2015

  50. [58]

    Hawthorne, F.; Raulino, P. R. E.; R., P. R.; Woellner, C. F. NOMAD Repository Dataset: Graphene MLIP dataset. Dataset, NOMAD Laboratory, 2025

    2025

  51. [59]

    , title =

    Hawthorne, F.; Raulino, P. R. E.; R., P. R.; Woellner, C. F. GrapheneMLIP. https://github.com/felipehgc/GrapheneMLIP, 2025 mcitethebibliography main.tex0000664000000000000000000010735115012172547011237 0ustar rootroot [journal=jctcce,manuscript=article,layout=twocolumn] achems...

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.