Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Technical Review
  • Published:

Roadmap on machine learning glassy dynamics

Nature Reviews Physics volume 7, pages 91–104 (2025)Cite this article

Subjects

Abstract

Unravelling the connections between microscopic structure, emergent physical properties and slow dynamics has long been a challenge when studying the glass transition. The absence of clear visible structural order in amorphous configurations complicates the identification of the key physical mechanisms underpinning slow dynamics. The difficulty in sampling equilibrated configurations at low temperatures hampers thorough numerical and theoretical investigations. We explore the potential of machine learning (ML) techniques to face these challenges, building on the algorithms that have revolutionized computer vision and image recognition. We present both successful ML applications and open problems for the future, such as transferability and interpretability of ML approaches. To foster a collaborative community effort, we also highlight the ‘GlassBench’ dataset, which provides simulation data and benchmarks for both 2D and 3D glass formers. We compare the performance of emerging ML methodologies, in line with benchmarking practices in image and text recognition. Our goal is to provide guidelines for the development of ML techniques in systems displaying slow dynamics and inspire new directions to improve our theoretical understanding of glassy liquids.

Key points

  • Systematic characterization of amorphous glassy structures can be addressed by unsupervised learning, which requires an adequate choice of structural descriptors.

  • Finding structure–dynamics relationships in glassy liquids is a task that has many analogies with image recognition and can be tackled using supervised learning with various neural network architectures already successful in image recognition.

  • Major challenges and potential breakthroughs await in transferring trained models to extremely low temperatures, using them to create ultrastable glasses and design new phenomenological glass models.

  • Future directions also encompass generative modelling of low-temperature equilibrium configurations and development of self-supervised and reinforcement learning approaches.

  • Publicly available datasets and unified benchmarks that are fundamental to stimulate further development of ML techniques in condensed matter physics are provided.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to the full article PDF.

USD 39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Visual summary of the scope of this Technical Review.
Fig. 2: Principal component analysis maps of the first two principal component projections \({\mathop{{\boldsymbol{X}}}\limits^{ \sim }}_{1}\) and \({\mathop{{\boldsymbol{X}}}\limits^{ \sim }}_{2}\) of the smooth bond-order (SBO) parameter in two representative systems.
Fig. 3: Typical supervised machine learning (ML) procedure in condensed matter.
Fig. 4: Training a model to predict single-particle propensity purely from structural properties.
Fig. 5: Training a transferable model to be accurate at different temperatures.
Fig. 6: Training a model that correctly predicts spatial dynamic heterogeneity.

Similar content being viewed by others

Data availability

The dataset GlassBench and Python scripts used to create the benchmarks presented in the section ‘Performance metrics and benchmarking’ are publicly available and can be downloaded from Zenodo at https://doi.org/10.5281/zenodo.10118191 (ref. 29).

References

  1. Ediger, M. D., Angell, C. A. & Nagel, S. R. Supercooled liquids and glasses. J. Phys. Chem. 100, 13200–13212 (1996).

    MATH  Google Scholar 

  2. Cavagna, A. Supercooled liquids for pedestrians. Phys. Rep. 476, 51–124 (2009).

    ADS  MATH  Google Scholar 

  3. LeCun, Y., Bengio, Y. & Hinton, G. Deep learning. Nature 521, 436–444 (2015).

    ADS  MATH  Google Scholar 

  4. Alzubaidi, L. et al. Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. J. Big Data 8, 1–74 (2021).

    MATH  Google Scholar 

  5. Royall, C. P. & Williams, S. R. The role of local structure in dynamical arrest. Phys. Rep. 560, 1–75 (2015).

    ADS  MathSciNet  MATH  Google Scholar 

  6. Tanaka, H., Tong, H., Shi, R. & Russo, J. Revealing key structural features hidden in liquids and glasses. Nat. Rev. Phys. 1, 333–348 (2019).

    MATH  Google Scholar 

  7. Marín-Aguilar, S., Wensink, H. H., Foffi, G. & Smallenburg, F. Tetrahedrality dictates dynamics in hard sphere mixtures. Phys. Rev. Lett. 124, 208005 (2020).

    ADS  Google Scholar 

  8. Richard, D. et al. Predicting plasticity in disordered solids from structural indicators. Phys. Rev. Mater. 4, 113609 (2020).

    Google Scholar 

  9. Boattini, E. et al. Autonomously revealing hidden local structures in supercooled liquids. Nat. Commun. 11, 5479 (2020).

    ADS  MATH  Google Scholar 

  10. Paret, J., Jack, R. L. & Coslovich, D. Assessing the structural heterogeneity of supercooled liquids through community inference. J. Chem. Phys. 152, 144502 (2020).

    ADS  MATH  Google Scholar 

  11. Oyama, N., Koyama, S. & Kawasaki, T. What do deep neural networks find in disordered structures of glasses? Front. Phys. 10, 1320 (2023).

    MATH  Google Scholar 

  12. Soltani, S., Sinclair, C. W. & Rottler, J. Exploring glassy dynamics with Markov state models from graph dynamical neural networks. Phys. Rev. E 106, 025308 (2022).

    ADS  MATH  Google Scholar 

  13. Widmer-Cooper, A. & Harrowell, P. Predicting the long-time dynamic heterogeneity in a supercooled liquid on the basis of short-time heterogeneities. Phys. Rev. Lett. 96, 185701 (2006).

    ADS  MATH  Google Scholar 

  14. Widmer-Cooper, A., Perry, H., Harrowell, P. & Reichman, D. R. Irreversible reorganization in a supercooled liquid originates from localized soft modes. Nat. Phys. 4, 711–715 (2008).

    Google Scholar 

  15. Berthier, L. & Biroli, G. Theoretical perspective on the glass transition and amorphous materials. Rev. Mod. Phys. 83, 587–645 (2011).

    ADS  MATH  Google Scholar 

  16. Schoenholz, S. S., Cubuk, E. D., Sussman, D. M., Kaxiras, E. & Liu, A. J. A structural approach to relaxation in glassy liquids. Nat. Phys. 12, 469–471 (2016).

    MATH  Google Scholar 

  17. Cubuk, E. D. et al. Identifying structural flow defects in disordered solids using machine-learning methods. Phys Rev. Lett. 114, 108001 (2015).

    ADS  MATH  Google Scholar 

  18. Bapst, V. et al. Unveiling the predictive power of static structure in glassy systems. Nat. Phys. 16, 448–454 (2020).

    MATH  Google Scholar 

  19. Yang, Z.-Y., Wei, D., Zaccone, A. & Wang, Y.-J. Machine-learning integrated glassy defect from an intricate configurational-thermodynamic-dynamic space. Phys. Rev. B 104, 064108 (2021).

    ADS  Google Scholar 

  20. Boattini, E., Smallenburg, F. & Filion, L. Averaging local structure to predict the dynamic propensity in supercooled liquids. Phys. Rev. Lett. 127, 088007 (2021).

    ADS  Google Scholar 

  21. Alkemade, R. M., Boattini, E., Filion, L. & Smallenburg, F. Comparing machine learning techniques for predicting glassy dynamics. J. Chem. Phys. 156, 204503 (2022).

    ADS  Google Scholar 

  22. Shiba, H., Hanai, M., Suzumura, T. & Shimokawabe, T. BOTAN: BOnd TArgeting Network for prediction of slow glassy dynamics by machine learning relative motion. J. Chem. Phys. 158, 084503 (2023).

    ADS  Google Scholar 

  23. Alkemade, R. M., Smallenburg, F. & Filion, L. Improving the prediction of glassy dynamics by pinpointing the local cage. J. Chem. Phys. 158, 134512 (2023).

    ADS  Google Scholar 

  24. Ciarella, S., Chiappini, M., Boattini, E., Dijkstra, M. & Janssen, L. M. C. Dynamics of supercooled liquids from static averaged quantities using machine learning. Mach. Learn. Sci. Technol. 4, 025010 (2023).

    ADS  Google Scholar 

  25. Pezzicoli, F. S., Charpiat, G. & Landes, F. P. Rotation-equivariant graph neural networks for learning glassy liquids representations. SciPost Phys. 16, 136 (2024).

    ADS  MathSciNet  Google Scholar 

  26. Ruiz-Garcia, M. et al. Discovering dynamic laws from observations: the case of self-propelled, interacting colloids. Phys. Rev. E 109, 064611 (2024).

    Google Scholar 

  27. Jung, G., Biroli, G. & Berthier, L. Predicting dynamic heterogeneity in glass-forming liquids by physics-inspired machine learning. Phys. Rev. Lett. 130, 238202 (2023).

    ADS  Google Scholar 

  28. Zhang, G. et al. Structuro-elasto-plasticity model for large deformation of disordered solids. Phys. Rev. Res. 4, 043026 (2022).

    MATH  Google Scholar 

  29. Jung, G. GlassBench. zenodo https://doi.org/10.5281/zenodo.10118191 (2023).

  30. Kob, W. & Andersen, H. C. Testing mode-coupling theory for a supercooled binary Lennard-Jones mixture I: the van Hove correlation function. Phys. Rev. E 51, 4626–4641 (1995).

    ADS  MATH  Google Scholar 

  31. Tarjus, G., Kivelson, D. & Viot, P. The viscous slowing down of supercooled liquids as a temperature-controlled super-Arrhenius activated process: a description in terms of frustration-limited domains. J. Phys. Condens. Matter 12, 6497 (2000).

    ADS  MATH  Google Scholar 

  32. Tanemura, M. et al. Geometrical analysis of crystallization of the soft-core model. Prog. Theor. Phys. 58, 1079–1095 (1977).

    ADS  MATH  Google Scholar 

  33. Malins, A., Williams, S. R., Eggers, J. & Royall, C. P. Identification of structure in condensed matter with the topological cluster classification. J. Chem. Phys. 139, 234506 (2013).

    ADS  MATH  Google Scholar 

  34. Honeycutt, J. D. & Andersen, H. C. Molecular dynamics study of melting and freezing of small Lennard-Jones clusters. J. Phys. Chem. 91, 4950–4963 (1987).

    MATH  Google Scholar 

  35. Lazar, E. A., Han, J. & Srolovitz, D. J. A topological framework for local structure analysis in condensed matter. Proc. Natl Acad. Sci. USA 112, E5769–E5776 (2015).

    ADS  MathSciNet  MATH  Google Scholar 

  36. Steinhardt, P. J., Nelson, D. R. & Ronchetti, M. Bond-orientational order in liquids and glasses. Phys. Rev. B 28, 784–805 (1983).

    ADS  MATH  Google Scholar 

  37. Mehta, P. et al. A high-bias, low-variance introduction to machine learning for physicists. Phys. Rep. 810, 1–124 (2019).

    ADS  MathSciNet  MATH  Google Scholar 

  38. Cheng, B. et al. Mapping materials and molecules. Acc. Chem. Res. 53, 1981–1991 (2020).

    MATH  Google Scholar 

  39. Glielmo, A. et al. Unsupervised learning methods for molecular simulation data. Chem. Rev. 121, 9722–9758 (2021).

    MATH  Google Scholar 

  40. Jarry, P. & Jakse, N. Medium range ordering in liquid Al-based alloys: towards a machine learning approach of solidification. IOP Conf. Ser. Mater. Sci. Eng. 1274, 012001 (2023).

    MATH  Google Scholar 

  41. Hu, W., Singh, R. R. P. & Scalettar, R. T. Discovering phases, phase transitions, and crossovers through unsupervised machine learning: a critical examination. Phys. Rev. E 95, 062122 (2017).

    ADS  Google Scholar 

  42. Rodriguez-Nieva, J. F. & Scheurer, M. S. Identifying topological order through unsupervised machine learning. Nat. Phys. 15, 790–795 (2019).

    MATH  Google Scholar 

  43. Mendes-Santos, T., Turkeshi, X., Dalmonte, M. & Rodriguez, A. Unsupervised learning universal critical behavior via the intrinsic dimension. Phys. Rev. X 11, 011040 (2021).

    Google Scholar 

  44. Bartók, A. P., Kondor, R. & Csányi, G. On representing chemical environments. Phys. Rev. B 87, 184115 (2013).

    ADS  MATH  Google Scholar 

  45. Parsaeifard, B. et al. An assessment of the structural resolution of various fingerprints commonly used in machine learning. Mach. Learn. Sci. Technol. 2, 015018 (2021).

    MATH  Google Scholar 

  46. Midtvedt, B. et al. Single-shot self-supervised object detection in microscopy. Nat. Commun. 13, 7492 (2022).

    ADS  MATH  Google Scholar 

  47. Caro, M. A., Deringer, V. L., Koskinen, J., Laurila, T. & Csányi, G. Growth mechanism and origin of high sp3 in tetrahedral amorphous carbon. Phys. Rev. Lett. 120, 166101 (2018).

    ADS  Google Scholar 

  48. Monserrat, B., Brandenburg, J. G., Engel, E. A. & Cheng, B. Liquid water contains the building blocks of diverse ice phases. Nat. Commun. 11, 5757 (2020).

    ADS  MATH  Google Scholar 

  49. Coslovich, D., Jack, R. L. & Paret, J. Dimensionality reduction of local structure in glassy binary mixtures. J. Chem. Phys. 157, 204503 (2022).

    MATH  Google Scholar 

  50. Banerjee, A., Hsu, H.-P., Kremer, K. & Kukharenko, O. Data-driven identification and analysis of the glass transition in polymer melts. ACS Macro Lett. 12, 679–684 (2023).

    MATH  Google Scholar 

  51. Banerjee, A., Iscen, A., Kremer, K. & Kukharenko, O. Determining glass transition in all-atom acrylic polymeric melt simulations using machine learning. J. Chem. Phys. 159, 074108 (2023).

    ADS  Google Scholar 

  52. Offei-Danso, A., Hassanali, A. & Rodriguez, A. High-dimensional fluctuations in liquid water: combining chemical intuition with unsupervised learning. J. Chem. Theory Comput. 18, 3136–3150 (2022).

    MATH  Google Scholar 

  53. Campadelli, P., Casiraghi, E., Ceruti, C. & Rozza, A. Intrinsic dimension estimation: relevant techniques and a benchmark framework. Math. Probl. Eng. 2015, e759567 (2015).

    MathSciNet  MATH  Google Scholar 

  54. Parsaeifard, B. & Goedecker, S. Manifolds of quasi-constant SOAP and ACSF fingerprints and the resulting failure to machine learn four-body interactions. J. Chem. Phys. 156, 034302 (2022).

    ADS  MATH  Google Scholar 

  55. Darby, J. P., Kermode, J. R. & Csányi, G. Compressing local atomic neighbourhood descriptors. npj Comput. Mater. 8, 1–13 (2022).

    Google Scholar 

  56. Darby, J. P. et al. Tensor-reduced atomic density representations. Phys. Rev. Lett. 131, 028001 (2023).

    ADS  Google Scholar 

  57. Coslovich, D., Ozawa, M. & Berthier, L. Local order and crystallization of dense polydisperse hard spheres. J. Phys. Condens. Matter 30, 144004 (2018).

    ADS  Google Scholar 

  58. Tong, H. & Tanaka, H. Emerging exotic compositional order on approaching low-temperature equilibrium glasses. Nat. Commun. 14, 4614 (2023).

    ADS  MATH  Google Scholar 

  59. Elliott, S. R. Medium-range structural order in covalent amorphous solids. Nature 354, 445–452 (1991).

    ADS  MATH  Google Scholar 

  60. Sheng, H., Luo, W., Alamgir, F., Bai, J. & Ma, E. Atomic packing and short-to-medium-range order in metallic glasses. Nature 439, 419–425 (2006).

    ADS  MATH  Google Scholar 

  61. Montes de Oca, J. M., Sciortino, F. & Appignanesi, G. A. A structural indicator for water built upon potential energy considerations. J. Chem. Phys. 152, 244503 (2020).

    ADS  MATH  Google Scholar 

  62. Faccio, C., Benzi, M., Zanetti-Polzi, L. & Daidone, I. Low- and high-density forms of liquid water revealed by a new medium-range order descriptor. J. Mol. Liq. 355, 118922 (2022).

    MATH  Google Scholar 

  63. Mauro, J. C., Tandia, A., Vargheese, K. D., Mauro, Y. Z. & Smedskjaer, M. M. Accelerating the design of functional glasses through modeling. Chem. Mater. 28, 4267–4277 (2016).

    MATH  Google Scholar 

  64. Cassar, D. R. et al. Predicting and interpreting oxide glass properties by machine learning using large datasets. Ceram. Int. 47, 23958–23972 (2021).

    MATH  Google Scholar 

  65. Bødker, M. L., Bauchy, M., Du, T., Mauro, J. C. & Smedskjaer, M. M. Predicting glass structure by physics-informed machine learning. npj Comput. Mater. 8, 192 (2022).

    ADS  Google Scholar 

  66. Bhattoo, R. et al. Artificial intelligence and machine learning in glass science and technology: 21 challenges for the 21st century. Int. J. Appl. Glass Sci. 12, 277–292 (2021).

    Google Scholar 

  67. Doliwa, B. & Heuer, A. What does the potential energy landscape tell us about the dynamics of supercooled liquids and glasses? Phys. Rev. Lett. 91, 235501 (2003).

    ADS  MATH  Google Scholar 

  68. Hocky, G. M., Coslovich, D., Ikeda, A. & Reichman, D. R. Correlation of local order with particle mobility in supercooled liquids is highly system dependent. Phys. Rev. Lett. 113, 157801 (2014).

    ADS  Google Scholar 

  69. Tong, H. & Tanaka, H. Revealing hidden structural order controlling both fast and slow glassy dynamics in supercooled liquids. Phys. Rev. X 8, 011041 (2018).

    Google Scholar 

  70. Schoenholz, S. S., Cubuk, E. D., Kaxiras, E. & Liu, A. J. Relationship between local structure and relaxation in out-of-equilibrium glassy systems. Proc. Natl Acad. Sci. USA 114, 263–267 (2017).

    ADS  MATH  Google Scholar 

  71. Sussman, D. M., Schoenholz, S. S., Cubuk, E. D. & Liu, A. J. Disconnecting structure and dynamics in glassy thin films. Proc. Natl Acad. Sci. USA 114, 10601–10605 (2017).

    ADS  MathSciNet  MATH  Google Scholar 

  72. Cubuk, E. D. et al. Structure-property relationships from universal signatures of plasticity in disordered solids. Science 358, 1033–1037 (2017).

    ADS  MATH  Google Scholar 

  73. Harrington, M., Liu, A. J. & Durian, D. J. Machine learning characterization of structural defects in amorphous packings of dimers and ellipses. Phys. Rev. E 99, 022903 (2019).

    ADS  MATH  Google Scholar 

  74. Ma, X. et al. Heterogeneous activation, local structure, and softness in supercooled colloidal liquids. Phys Rev. Lett.122, 028001 (2019).

    ADS  MATH  Google Scholar 

  75. Cubuk, E. D., Liu, A. J., Kaxiras, E. & Schoenholz, S. S. Unifying framework for strong and fragile liquids via machine learning: a study of liquid silica. Preprint at https://doi.org/10.48550/arXiv.2008.09681 (2020).

  76. Ridout, S. A., Rocks, J. W. & Liu, A. J. Correlation of plastic events with local structure in jammed packings across spatial dimensions. Proc. Natl Acad. Sci. USA 119, e2119006119 (2022).

    MATH  Google Scholar 

  77. Tah, I., Ridout, S. A., & Liu, A. J. Fragility in glassy liquids: a structural approach based on machine learning. J. Chem. Phys.157, 124501 (2022).

    ADS  MATH  Google Scholar 

  78. Liu, H., Smedskjaer, M. M. & Bauchy, M. Deciphering a structural signature of glass dynamics by machine learning. Phys. Rev. B 106, 214206 (2022).

    ADS  Google Scholar 

  79. Zhang, G., Ridout, S. A. & Liu, A. J. Interplay of rearrangements, strain, and local structure during avalanche propagation. Phys. Rev. X 11, 041019 (2021).

    Google Scholar 

  80. Xiao, H. et al. Machine learning-informed structuro-elastoplasticity predicts ductility of disordered solids. Preprint at https://doi.org/10.48550/arXiv.2303.12486 (2023).

  81. Widmer-Cooper, A., Harrowell, P. & Fynewever, H. How reproducible are dynamic heterogeneities in a supercooled liquid? Phys. Rev. Lett. 93, 135701 (2004).

    ADS  Google Scholar 

  82. Berthier, L. & Jack, R. L. Structure and dynamics of glass formers: predictability at large length scales. Phys. Rev. E 76, 041509 (2007).

    ADS  MATH  Google Scholar 

  83. Jung, G., Biroli, G. & Berthier, L. Dynamic heterogeneity at the experimental glass transition predicted by transferable machine learning. Phys. Rev. B 109, 064205 (2024).

    ADS  Google Scholar 

  84. Berthier, L., Biroli, G., Bouchaud, J.-P., Cipelletti, L. & Saarloos, W. Dynamical Heterogeneities in Glasses, Colloids, and Granular Media (Oxford Univ. Press, 2011).

  85. Gidaris, S., Singh, P. & Komodakis, N. Unsupervised representation learning by predicting image rotations. Preprint at https://doi.org/10.48550/arXiv.1803.07728 (2018).

  86. Toninelli, C., Wyart, M., Berthier, L., Biroli, G. & Bouchaud, J.-P. Dynamical susceptibility of glass formers: contrasting the predictions of theoretical scenarios. Phys. Rev. E 71, 041505 (2005).

    ADS  MATH  Google Scholar 

  87. Kingma, D. P. et al. An introduction to variational autoencoders. Found. Trends Mach. Learn. 12, 307–392 (2019).

    MATH  Google Scholar 

  88. Wang, Q. & Zhang, L. Inverse design of glass structure with deep graph neural networks. Nat. Commun. 12, 5359 (2021).

    ADS  MATH  Google Scholar 

  89. Kivelson, S. & Kivelson, S. Understanding complexity. Nat. Phys. 14, 426–427 (2018).

    MATH  Google Scholar 

  90. Wang, Q. et al. Predicting the propensity for thermally activated β events in metallic glasses via interpretable machine learning. npj Comput. Mater. 6, 194 (2020).

    ADS  MATH  Google Scholar 

  91. Miao, S., Liu, M. & Li, P. Interpretable and generalizable graph learning via stochastic attention mechanism. In Proc. 39th International Conference on Machine Learning, Volume 162 of Proceedings of Machine Learning Research (eds Chaudhuri, K. et al.) 15524–15543 (PMLR, 2022).

  92. Duede, E. Deep learning opacity in scientific discovery. Philos. Sci. 90, 1089–1099 (2023).

    MathSciNet  MATH  Google Scholar 

  93. Glielmo, A., Zeni, C., Cheng, B., Csányi, G. & Laio, A. Ranking the information content of distance measures. PNAS Nexus 1, pgac039 (2022).

    Google Scholar 

  94. Sandberg, J., Voigtmann, T., Devijver, E. & Jakse, N. Feature selection for high-dimensional neural network potentials with the adaptive group lasso. Mach. Learn. Sci. Technol. 5, 025043 (2024).

    ADS  Google Scholar 

  95. Sharma, A., Liu, C. & Ozawa, M. Selecting relevant structural features for glassy dynamics by information imbalance. J. Chem. Phys. 161, 184506 (2024).

    Google Scholar 

  96. Berthier, L., Flenner, E. & Szamel, G. Glassy dynamics in dense systems of active particles. J. Chem. Phys. 150, 200901 (2019).

    ADS  MATH  Google Scholar 

  97. Janzen, G. & Janssen, L. M. C. Rejuvenation and memory effects in active glasses induced by thermal and active cycling. Phys. Rev. Res. 6, 023257 (2024).

    MATH  Google Scholar 

  98. Janzen, G. et al. Dead or alive: distinguishing active from passive particles using supervised learning. Europhys. Lett. 143, 17004 (2023).

    ADS  Google Scholar 

  99. Janzen, G. et al. Classifying the age of a glass based on structural properties: a machine learning approach. Phys. Rev. Mater. 8, 025602 (2024).

    MATH  Google Scholar 

  100. Scalliet, C., Guiselin, B. & Berthier, L. Excess wings and asymmetric relaxation spectra in a facilitated trap model. J. Chem. Phys. 155, 064505 (2021).

    ADS  Google Scholar 

  101. Guiselin, B., Scalliet, C. & Berthier, L. Microscopic origin of excess wings in relaxation spectra of supercooled liquids. Nat. Phys. 18, 468–472 (2022).

    MATH  Google Scholar 

  102. Nicolas, A., Ferrero, E. E., Martens, K. & Barrat, J.-L. Deformation and flow of amorphous solids: Insights from elastoplastic models. Rev. Mod. Phys. 90, 045006 (2018).

    ADS  MATH  Google Scholar 

  103. Ozawa, M. & Biroli, G. Elasticity, facilitation and dynamic heterogeneity in glass-forming liquids. Phys. Rev. Lett. 130, 138201 (2023).

    ADS  MATH  Google Scholar 

  104. Tahaei, A., Biroli, G., Ozawa, M., Popović, M. & Wyart, M. Scaling description of dynamical heterogeneity and avalanches of relaxation in glass-forming liquids. Phys. Rev. X 13, 031034 (2023).

    Google Scholar 

  105. Lerbinger, M., Barbot, A., Vandembroucq, D. & Patinet, S. Relevance of shear transformations in the relaxation of supercooled liquids. Phys. Rev. Lett. 129, 195501 (2022).

    ADS  MATH  Google Scholar 

  106. Chacko, R. N. et al. Elastoplasticity mediates dynamical heterogeneity below the mode coupling temperature. Phys Rev. Lett. 127, 048002 (2021).

    ADS  MATH  Google Scholar 

  107. Monthus, C. & Bouchaud, J. P. Models of traps and glass phenomenology. J. Phys. A Math. Gen. 29 3847 (1996).

    ADS  MATH  Google Scholar 

  108. Ridout, S. A., Tah, I. & Liu, A. J. Building a “trap model” of glassy dynamics from a local structural predictor of rearrangements. Europhys. Lett. 144, 47001 (2023).

    ADS  Google Scholar 

  109. Ridout, S. A. & Liu, A. J. The dynamics of machine-learned “softness” in supercooled liquids describe dynamical heterogeneity. Preprint at https://doi.org/10.48550/arXiv.2406.05868 (2024).

  110. Ciarella, S. et al. Finding defects in glasses through machine learning. Nat. Commun. 14, 4229 (2023).

    ADS  MATH  Google Scholar 

  111. Richard, D., Kapteijns, G. & Lerner, E. Detecting low-energy quasilocalized excitations in computer glasses. Phys. Rev. E 108, 044124 (2023).

    ADS  Google Scholar 

  112. Scalliet, C., Guiselin, B. & Berthier, L. Thirty milliseconds in the life of a supercooled liquid. Phys. Rev. X 12, 041028 (2022).

    MATH  Google Scholar 

  113. Ediger, M. D. Spatially heterogeneous dynamics in supercooled liquids. Annu. Rev. Phys. Chem. 51, 99–128 (2000).

    ADS  MATH  Google Scholar 

  114. Tong, H. & Tanaka, H. Structural order as a genuine control parameter of dynamics in simple glass formers. Nat. Commun. 10, 5596 (2019).

    ADS  MATH  Google Scholar 

  115. Lačević, N., Starr, F. W., Schrøder, T. B. & Glotzer, S. C. Spatially heterogeneous dynamics investigated via a time-dependent four-point density correlation function. J. Chem. Phys. 119, 7372–7387 (2003).

    ADS  MATH  Google Scholar 

  116. Flenner, E., Zhang, M. & Szamel, G. Analysis of a growing dynamic length scale in a glass-forming binary hard-sphere mixture. Phys. Rev. E 83, 051501 (2011).

    ADS  MATH  Google Scholar 

  117. Jiang, X., Tian, Z., Li, K. & Hu, W. A geometry-enhanced graph neural network for learning the smoothness of glassy dynamics from static structure. J. Chem. Phys. 159, 144504 (2023).

    ADS  MATH  Google Scholar 

  118. Swendsen, R. H. & Wang, J.-S. Replica Monte Carlo simulation of spin-glasses. Phys. Rev. Lett. 57, 2607 (1986).

    ADS  MathSciNet  MATH  Google Scholar 

  119. Berthier, L., Coslovich, D., Ninarello, A. & Ozawa, M. Equilibrium sampling of hard spheres up to the jamming density and beyond. Phys. Rev. Lett. 116, 238002 (2016).

    ADS  MathSciNet  Google Scholar 

  120. Fan, Z. & Ma, E. Predicting orientation-dependent plastic susceptibility from static structure in amorphous solids via deep learning. Nat. Commun. 12, 1506 (2021).

    ADS  MATH  Google Scholar 

  121. Du, T. et al. Predicting fracture propensity in amorphous alumina from its static structure using machine learning. ACS Nano 15, 17705–17716 (2021).

    Google Scholar 

  122. Font-Clos, F. et al. Predicting the failure of two-dimensional silica glasses. Nat. Commun. 13, 2820 (2022).

    ADS  MATH  Google Scholar 

  123. Liu, H., Fu, Z., Yang, K., Xu, X. & Bauchy, M. Machine learning for glass science and engineering: a review. J. Non Cryst. Solids 557, 119419 (2021).

    Google Scholar 

  124. Cassar, D. R. GlassNet: a multitask deep neural network for predicting many glass properties. Ceram. Int. 49, 36013–36024 (2023).

    MATH  Google Scholar 

  125. Tandia, A., Onbasli, M. C. & Mauro, J. C. in Springer Handbook of Glass 1157–1192 (2019).

  126. Merchant, A. et al. Scaling deep learning for materials discovery. Nature 624, 80–85 (2023).

    ADS  MATH  Google Scholar 

  127. Unke, O. T. et al. Machine learning force fields. Chem. Rev. 121, 10142–10186 (2021).

    Google Scholar 

  128. Volpe, G. et al. Roadmap on deep learning for microscopy. Preprint at https://doi.org/10.48550/arXiv.2303.03793 (2023).

  129. Midtvedt, B., Pineda, J., Klein Morberg, H., Manzo, C. & Volpe, G. DeepTrack2. https://github.com/softmatterlab/DeepTrack2 (2024).

  130. Gabrié, M. Mean-field inference methods for neural networks. J. Phys. A Math. Theor. 53, 223002 (2020).

    ADS  MathSciNet  MATH  Google Scholar 

  131. Merchant, A., Metz, L., Schoenholz, S. S. & Cubuk, E. D. Learn2hop: learned optimization on rough landscapes. In International Conference on Machine Learning 7643–7653 (PMLR, 2021).

  132. Gabrié, M., Ganguli, S., Lucibello, C. & Zecchina, R. Neural networks: from the perceptron to deep nets. Preprint at https://doi.org/10.48550/arXiv.2304.06636 (2023).

  133. Bonnaire, T. et al. High-dimensional non-convex landscapes and gradient descent dynamics. J. Stat. Mech. 104004 (2024).

  134. Mézard, M. Spin glass theory and its new challenge: structured disorder. Indian J. Phys. 98, 3757 (2023).

    MATH  Google Scholar 

  135. Vaswani, A. Attention is all you need. In 31st Conference on Neural Information Processing Systems (NIPS, 2017).

  136. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).

    ADS  MATH  Google Scholar 

  137. Baek, M. et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science 373, 871–876 (2021).

    ADS  MATH  Google Scholar 

  138. Bratholm, L. A. et al. A community-powered search of machine learning strategy space to find NMR property prediction models. PLoS ONE 16, e0253612 (2021).

    Google Scholar 

  139. Qin, Y. et al. A dual-stage attention-based recurrent neural network for time series prediction. Preprint at https://doi.org/10.48550/arXiv.1704.02971 (2017).

  140. Chapelle, O., Schölkopf, B. & Zien, A. (eds.) Semi-Supervised Learning (MIT Press, 2006).

  141. Rong, Y. et al. Self-supervised graph transformer on large-scale molecular data. Adv. Neural Inf. Process. Syst. 33, 12559–12571 (2020).

    MATH  Google Scholar 

  142. Magar, R., Wang, Y. & Barati Farimani, A. Crystal twins: self-supervised learning for crystalline material property prediction. npj Comput. Mater. 8, 231 (2022).

    ADS  MATH  Google Scholar 

  143. Zhang, Z. et al. Graph self-supervised learning for optoelectronic properties of organic semiconductors. In ICML 2022 2nd AI for Science Workshop (2022).

  144. Kaelbling, L. P., Littman, M. L. & Moore, A. W. Reinforcement learning: a survey. J. Artif. Intell. Res. 4, 237–285 (1996).

    MATH  Google Scholar 

  145. Shin, K. et al. Enhancing biomolecular sampling with reinforcement learning: a tree search molecular dynamics simulation method. ACS Omega 4, 138530–13862 (2019).

    Google Scholar 

  146. Fan, C. et al. Searching for spin glass ground states through deep reinforcement learning. Nat. Commun. 14, 725 (2023).

    ADS  MATH  Google Scholar 

  147. Ahuja, K., Green, W. H. & Li, Y.-P. Learning to optimize molecular geometries using reinforcement learning. J. Chem. Theory Comput. 17, 818–825 (2021).

    MATH  Google Scholar 

  148. Bihani, V., Manchanda, S., Sastry, S., Ranu, S. & Krishnan, N. A. Stridernet: a graph reinforcement learning approach to optimize atomic structures on rough energy landscapes. In International Conference on Machine Learning 2431–2451 (PMLR, 2023).

  149. Bojesen, T. A. Policy-guided Monte Carlo: reinforcement-learning Markov chain dynamics. Phys. Rev. E 98, 063303 (2018).

    ADS  MATH  Google Scholar 

  150. Galliano, L., Rende, R. & Coslovich, D. Policy-guided Monte Carlo on general state spaces: application to glass-forming mixtures. J. Chem. Phys. 161, 064503 (2024).

    Google Scholar 

  151. Christiansen, H., Errica, F. & Alesiani, F. Self-tuning Hamiltonian Monte Carlo for accelerated sampling. J. Chem. Phys. 159, 234109 (2023).

    ADS  Google Scholar 

  152. Gabrié, M., Rotskoff, G. M. & Vanden-Eijnden, E. Adaptive Monte Carlo augmented with normalizing flows. Proc. Natl Acad. Sci. USA 119, e2109420119 (2022).

    MathSciNet  MATH  Google Scholar 

  153. Ninarello, A., Berthier, L. & Coslovich, D. Models and algorithms for the next generation of glass transition studies. Phys. Rev. X 7, 021039 (2017).

    Google Scholar 

  154. Berthier, L. & Reichman, D. R. Modern computational studies of the glass transition. Nat. Rev. Phys. 5, 102–116 (2023).

    MATH  Google Scholar 

  155. Noé, F., Olsson, S., Köhler, J. & Wu, H. Boltzmann generators: sampling equilibrium states of many-body systems with deep learning. Science 365, eaaw1147 (2019).

    ADS  MATH  Google Scholar 

  156. Wu, D., Wang, L. & Zhang, P. Solving statistical mechanics using variational autoregressive networks. Phys. Rev. Lett. 122, 080602 (2019).

    ADS  MATH  Google Scholar 

  157. Köhler, J., Klein, L. & Noé, F. Equivariant flows: exact likelihood generative learning for symmetric densities. In International Conference on Machine Learning 5361–5370 (PMLR, 2020).

  158. Dibak, M., Klein, L., Krämer, A. & Noé, F. Temperature steerable flows and Boltzmann generators. Phys. Rev. Res. 4, L042005 (2022).

    Google Scholar 

  159. Invernizzi, M., Krämer, A., Clementi, C. & Noé, F. Skipping the replica exchange ladder with normalizing flows. J. Phys. Chem. Lett. 13, 11643–11649 (2022).

    Google Scholar 

  160. Xu, M. et al. Geodiff: a geometric diffusion model for molecular conformation generation. Preprint at https://doi.org/10.48550/arXiv.2203.02923 (2022).

  161. Coretti, A., Falkner, S., Geissler, P. & Dellago, C. Learning mappings between equilibrium states of liquid systems using normalizing flows. Preprint at https://doi.org/10.48550/arXiv.2208.10420 (2022).

  162. van Leeuwen, S., de Alba Ortíz, A. P. & Dijkstra, M. A Boltzmann generator for the isobaric-isothermal ensemble. Preprint at https://doi.org/10.48550/arXiv.2305.08483 (2023).

  163. Jung, G., Biroli, G. & Berthier, L. Normalizing flows as an enhanced sampling method for atomistic supercooled liquids. Mach. Learn. Sci. Technol. 5, 035053 (2024).

    Google Scholar 

  164. McNaughton, B., Milošević, M., Perali, A. & Pilati, S. Boosting Monte Carlo simulations of spin glasses using autoregressive neural networks. Phys. Rev. E 101, 053312 (2020).

    ADS  Google Scholar 

  165. Hibat-Allah, M., Inack, E. M., Wiersema, R., Melko, R. G. & Carrasquilla, J. Variational neural annealing. Nat. Mach. Intell. 3, 952–961 (2021).

    MATH  Google Scholar 

  166. Wu, D., Rossi, R. & Carleo, G. Unbiased Monte Carlo cluster updates with autoregressive neural networks. Phys. Rev. Res. 3, L042024 (2021).

    MATH  Google Scholar 

  167. Inack, E. M., Morawetz, S. & Melko, R. G. Neural annealing and visualization of autoregressive neural networks in the Newman–Moore model. Condens. Matter 7, 38 (2022).

    Google Scholar 

  168. Ciarella, S., Trinquier, J., Weigt, M. & Zamponi, F. Machine-learning-assisted Monte Carlo fails at sampling computationally hard problems. Mach. Learn. Sci. Technol. 4, 010501 (2023).

    ADS  Google Scholar 

  169. Schuetz, M. J., Brubaker, J. K., Zhu, Z. & Katzgraber, H. G. Graph coloring with physics-inspired graph neural networks. Phys. Rev. Res. 4, 043131 (2022).

    Google Scholar 

  170. Schuetz, M. J., Brubaker, J. K. & Katzgraber, H. G. Combinatorial optimization with physics-inspired graph neural networks. Nat. Mach. Intell. 4, 367–377 (2022).

    Google Scholar 

  171. Albergo, M. S., Kanwar, G. & Shanahan, P. E. Flow-based generative models for Markov chain Monte Carlo in lattice field theory. Phys. Rev. D 100, 034515 (2019).

    ADS  MathSciNet  Google Scholar 

  172. Kanwar, G. et al. Equivariant flow-based sampling for lattice gauge theory. Phys. Rev. Lett. 125, 121601 (2020).

    ADS  MathSciNet  MATH  Google Scholar 

  173. de Haan, P., Rainone, C., Cheng, M. C. & Bondesan, R. Scaling up machine learning for quantum field theory with equivariant continuous flows. Preprint at https://doi.org/10.48550/arXiv.2110.02673 (2021).

  174. Gerdes, M., de Haan, P., Rainone, C., Bondesan, R. & Cheng, M. C. Learning lattice quantum field theories with equivariant continuous flows. SciPost Phys. 15, 238 (2023).

    ADS  MathSciNet  MATH  Google Scholar 

  175. Luo, D., Carleo, G., Clark, B. K. & Stokes, J. Gauge equivariant neural networks for quantum lattice gauge theories. Phys. Rev. Lett. 127, 276402 (2021).

    ADS  MathSciNet  MATH  Google Scholar 

  176. Marchand, T., Ozawa, M., Biroli, G. & Mallat, S. Wavelet conditional renormalization group. Preprint at https://doi.org/10.48550/arXiv.2207.04941 (2022).

  177. Angelini, M. C. & Ricci-Tersenghi, F. Modern graph neural networks do worse than classical greedy algorithms in solving combinatorial optimization problems like maximum independent set. Nat. Mach. Intell. 5, 29–31 (2023).

    MATH  Google Scholar 

  178. Boettcher, S. Inability of a graph neural network heuristic to outperform greedy algorithms in solving combinatorial optimization problems. Nat. Mach. Intell. 5, 24–25 (2023).

    MATH  Google Scholar 

  179. Boettcher, S. Deep reinforced learning heuristic tested on spin-glass ground states: the larger picture. Nat. Commun. 14, 5658 (2023).

    ADS  MATH  Google Scholar 

  180. Ghio, D., Dandi, Y., Krzakala, F. & Zdeborová, L. Sampling with flows, diffusion and autoregressive neural networks: a spin-glass perspective. Proc. Natl Acad. Sci. USA 121, e2311810121 (2024).

    MATH  Google Scholar 

Download references

Acknowledgements

This paper originates from discussions and interactions at the AISSAI (AI for science, science for AI) workshop on ‘Machine Learning Glasses’ held in November 2022 in Paris. This workshop was organized by G.B., L.B. and G.J. The authors thank all participants for their attendance, discussions and feedback, in particular A. Banerjee, L. Janssen, M. Ruiz Garcia, S. Patinet, C. Scalliet, D. Richard, J. Rottler, O. Dauchot and O. Kukharenko for their valuable contributions. F.S.P. is supported by a public grant overseen by the French National Research Agency (ANR) through the programme UDOPIA, project funded by the ANR-20-THIA-0013-01. F.S.P. was granted access to the HPC resources of IDRIS under the allocation 2022-AD011014066 made by GENCI. H.S. acknowledges computational resources provided by ‘Joint Usage/Research Center for Interdisciplinary Large-scale Information Infrastructures (JHPCN)’ and ‘High Performance Computing Infrastructure (HPCI)’ in Japan (project ID: jh230064). A.J.L. is supported by the Simons Foundation via the Investigator Award #327939. In addition, A.J.L. thanks CCB at the Flatiron Institute and the Isaac Newton Institute for Mathematical Sciences under the programme ‘New Statistical Physics in Living Matter’ (EPSRC grant EP/R014601/1) for the support and hospitality. This work was supported by a grant from the Simons Foundation (#454933 to L.B., #454935 to G.B.). G.B. acknowledges funding from the French government under the management of Agence Nationale de la Recherche as part of the ‘Investissements d’avenir’ programme, reference ANR-19-P3IA-0001 (PRAIRIE 3IA Institute).

Author information

Authors and Affiliations

  1. Laboratoire Charles Coulomb (L2C), Université de Montpellier, CNRS, Montpellier, France

    Gerhard Jung & Ludovic Berthier

  2. Laboratoire Interdisciplinaire de Physique (LIPhy), Université Grenoble Alpes, Saint-Martin-d’Hères, France

    Gerhard Jung

  3. Soft Condensed Matter and Biophysics, Debye Institute for Nanomaterials Science, Utrecht University, Utrecht, The Netherlands

    Rinske M. Alkemade & Laura Filion

  4. GoogleDeepMind, London, UK

    Victor Bapst

  5. Dipartimento di Fisica, Università di Trieste, Trieste, Italy

    Daniele Coslovich

  6. TAU Team, Laboratoire Interdisciplinaire des Sciences du Numérique, CNRS, INRIA, Université Paris-Saclay, Gif-sur-Yvette, France

    François P. Landes & Francesco Saverio Pezzicoli

  7. Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, PA, USA

    Andrea J. Liu

  8. Santa Fe Institute, Santa Fe, NM, USA

    Andrea J. Liu

  9. Graduate School of Information Science, University of Hyogo, Kobe, Japan

    Hayato Shiba

  10. Department of Physics, University of Gothenburg, Gothenburg, Sweden

    Giovanni Volpe

  11. Dipartimento di Fisica, Sapienza Università di Roma, Rome, Italy

    Francesco Zamponi

  12. Gulliver, UMR CNRS 7083, ESPCI Paris-PSL Research University, Paris, France

    Ludovic Berthier

  13. Laboratoire de Physique de l’Ecole Normale Supérieure, ENS Université PSL, CNRS, Sorbonne Université, Université de Paris, Paris, France

    Giulio Biroli

Authors
  1. Gerhard Jung
    View author publications

    Search author on:PubMed Google Scholar

  2. Rinske M. Alkemade
    View author publications

    Search author on:PubMed Google Scholar

  3. Victor Bapst
    View author publications

    Search author on:PubMed Google Scholar

  4. Daniele Coslovich
    View author publications

    Search author on:PubMed Google Scholar

  5. Laura Filion
    View author publications

    Search author on:PubMed Google Scholar

  6. François P. Landes
    View author publications

    Search author on:PubMed Google Scholar

  7. Andrea J. Liu
    View author publications

    Search author on:PubMed Google Scholar

  8. Francesco Saverio Pezzicoli
    View author publications

    Search author on:PubMed Google Scholar

  9. Hayato Shiba
    View author publications

    Search author on:PubMed Google Scholar

  10. Giovanni Volpe
    View author publications

    Search author on:PubMed Google Scholar

  11. Francesco Zamponi
    View author publications

    Search author on:PubMed Google Scholar

  12. Ludovic Berthier
    View author publications

    Search author on:PubMed Google Scholar

  13. Giulio Biroli
    View author publications

    Search author on:PubMed Google Scholar

Contributions

G.J., L.B. and G.B. coordinated the manuscript and submission. D.C., L.F., A.J.L. and G.J. drafted original versions of the sections from ‘Machine learning locally favoured structures’ to ‘Performance metrics and benchmarking’. V.B., G.V., F.Z. and F.P.L. drafted original versions of the ‘Outlook’ section. L.F., R.M.A., F.P.L., F.S.P., D.C., H.S. and G.J. devised the benchmarks. All authors participated in proofreading and corrections.

Corresponding author

Correspondence to Ludovic Berthier.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Reviews Physics thanks Han Liu and the other, anonymous, referee(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Jung, G., Alkemade, R.M., Bapst, V. et al. Roadmap on machine learning glassy dynamics. Nat Rev Phys 7, 91–104 (2025). https://doi.org/10.1038/s42254-024-00791-4

Download citation

  • Accepted:

  • Published:

  • Version of record:

  • Issue date:

  • DOI: https://doi.org/10.1038/s42254-024-00791-4

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing