Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
97 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
5 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Predicting and Interpreting Energy Barriers of Metallic Glasses with Graph Neural Networks (2401.08627v3)

Published 8 Dec 2023 in cond-mat.dis-nn, cond-mat.mtrl-sci, and cs.LG

Abstract: Metallic Glasses (MGs) are widely used materials that are stronger than steel while being shapeable as plastic. While understanding the structure-property relationship of MGs remains a challenge in materials science, studying their energy barriers (EBs) as an intermediary step shows promise. In this work, we utilize Graph Neural Networks (GNNs) to model MGs and study EBs. We contribute a new dataset for EB prediction and a novel Symmetrized GNN (SymGNN) model that is E(3)-invariant in expectation. SymGNN handles invariance by aggregating over orthogonal transformations of the graph structure. When applied to EB prediction, SymGNN are more accurate than molecular dynamics (MD) local-sampling methods and other machine-learning models. Compared to precise MD simulations, SymGNN reduces the inference time on new MGs from roughly 41 days to less than one second. We apply explanation algorithms to reveal the relationship between structures and EBs. The structures that we identify through explanations match the medium-range order (MRO) hypothesis and possess unique topological properties. Our work enables effective prediction and interpretation of MG EBs, bolstering material science research.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (39)
  1. Unveiling the predictive power of static structure in glassy systems. Nature Physics, 16(4):448–454, April 2020.
  2. GT Barkema and Normand Mousseau. Event-based relaxation of continuous disordered systems. Physical review letters, 77(21):4358, 1996.
  3. Mace: Higher order equivariant message passing neural networks for fast and accurate force fields. Advances in Neural Information Processing Systems, 35:11423–11436, 2022.
  4. E (3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials. Nature communications, 13(1):2453, 2022.
  5. Some improvements of the activation-relaxation technique method for finding transition pathways on potential energy surfaces. The Journal of chemical physics, 130(11), 2009.
  6. Structural and topological nature of plasticity in sheared granular materials. Nature communications, 9(1):2911, 2018.
  7. Structure-property relationships from universal signatures of plasticity in disordered solids. Science, 358(6366):1033–1037, 2017.
  8. Supercooled liquids and the glass transition. Nature, 410(6825):259–267, 2001.
  9. Soft spots and their structural signature in a metallic glass. Proceedings of the National Academy of Sciences, 111(39):14052–14056, 2014.
  10. How thermally activated deformation starts in metallic glass. Nature communications, 5(1):5083, 2014.
  11. Energy landscape-driven non-equilibrium evolution of inherent structure in disordered material. Nature communications, 8(1):15417, 2017.
  12. Inductive representation learning on large graphs, 2018.
  13. William G Hoover. Canonical dynamics: Equilibrium phase-space distributions. Physical review A, 31(3):1695, 1985.
  14. Rossmann-toolbox: a deep learning-based protocol for the prediction and design of cofactor specificity in Rossmann fold proteins. Briefings in Bioinformatics, 23(1):bbab371, 09 2021.
  15. Semi-supervised classification with graph convolutional networks, 2017.
  16. Equiformer: Equivariant graph attention transformer for 3d atomistic graphs. arXiv preprint arXiv:2206.11990, 2022.
  17. Orphicx: A causality-inspired latent variable model for interpreting graph neural networks, 2022.
  18. A unified approach to interpreting model predictions, 2017.
  19. Parameterized explainer for graph neural network, 2020.
  20. KV Mardia and PJ Zemroch. Algorithm as 86: The von mises distribution function. Journal of the Royal Statistical Society. Series C (Applied Statistics), 24(2):268–272, 1975.
  21. Development of suitable interatomic potentials for simulation of liquid and amorphous cu–zr alloys. Philosophical Magazine, 89(11):967–987, 2009.
  22. Shuichi Nosé. A unified formulation of the constant temperature molecular dynamics methods. The Journal of chemical physics, 81(1):511–519, 1984.
  23. Connecting local yield stresses with plastic activity in amorphous solids. Physical review letters, 117(4):045501, 2016.
  24. Steve Plimpton. Fast parallel algorithms for short-range molecular dynamics. Journal of computational physics, 117(1):1–19, 1995.
  25. On the convergence of adam and beyond. arXiv preprint arXiv:1904.09237, 2019.
  26. "why should i trust you?": Explaining the predictions of any classifier, 2016.
  27. E(n) equivariant graph neural networks, 2022.
  28. Gregory G Slabaugh. Computing euler angles from a rotation matrix. Retrieved on August, 6(2000):39–63, 1999.
  29. What do we learn from the local geometry of glass-forming liquids? Physical review letters, 89(12):125501, 2002.
  30. The energy landscape governs ductility in disordered materials. Materials Horizons, 8(4):1242–1252, 2021.
  31. Bulk metallic glasses’ response to oscillatory stress is governed by the topography of the energy landscape. The Journal of Physical Chemistry B, 124(49):11294–11298, 2020.
  32. Graph attention networks. arXiv preprint arXiv:1710.10903, 2017.
  33. Predicting the propensity for thermally activated β𝛽\betaitalic_β events in metallic glasses via interpretable machine learning, 2020.
  34. Predicting shear transformation events in metallic glasses. Physical review letters, 120(12):125503, 2018.
  35. Gnnexplainer: Generating explanations for graph neural networks, 2019.
  36. Correlation between β𝛽\betaitalic_β relaxation and self-diffusion of the smallest constituting atoms in metallic glasses. Physical review letters, 109(9):095508, 2012.
  37. On explainability of graph neural networks via subgraph explorations, 2021.
  38. Regexplainer: Generating explanations for graph neural networks in regression task. arXiv preprint arXiv:2307.07840, 2023.
  39. Gstarx: Explaining graph neural networks with structure-aware cooperative games, 2022.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (5)
  1. Haoyu Li (56 papers)
  2. Shichang Zhang (21 papers)
  3. Longwen Tang (1 paper)
  4. Mathieu Bauchy (32 papers)
  5. Yizhou Sun (149 papers)

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets