Towards exact molecular dynamics simulations with machine-learned force fields

Abstract

Molecular dynamics (MD) simulations employing classical force fields constitute the cornerstone of contemporary atomistic modeling in chemistry, biology, and materials science. However, the predictive power of these simulations is only as good as the underlying interatomic potential. Classical potentials often fail to faithfully capture key quantum effects in molecules and materials. Here we enable the direct construction of flexible molecular force fields from high-level ab initio calculations by incorporating spatial and temporal physical symmetries into a gradient-domain machine learning (sGDML) model in an automatic data-driven way. The developed sGDML approach faithfully reproduces global force fields at quantum-chemical CCSD(T) level of accuracy and allows converged molecular dynamics simulations with fully quantized electrons and nuclei. We present MD simulations, for flexible molecules with up to a few dozen atoms and provide insights into the dynamical behavior of these molecules. Our approach provides the key missing ingredient for achieving spectroscopic accuracy in molecular simulations.

Full text

Its licence does not allow us to show the full text. Read it at PubMed Central:

References 70

1. 1.

   Tuckerman, M. Statistical Mechanics: Theory and Molecular Simulation (Oxford University Press, Oxford, UK, 2010).

   Not in our library

2. 2.

   Koch, W. & Holthausen, M. C. A Chemist's Guide to Density Functional Theory (John Wiley & Sons, Hoboken, New Jersey, USA, 2015).

   Not in our library

3. 3.

   Partridge H, Schwenke DW. The determination of an accurate isotope dependent potential energy surface for water from extensive ab initio calculations and experimental data. J. Chem. Phys. 1997;106:4618-4639.

   Not in our library10.1063/1.473987

4. 4.

   Mizukami W, Habershon S, Tew DP. A compact and accurate semi-global potential energy surface for malonaldehyde from constrained least squares regression. J. Chem. Phys. 2014;141:144310.

   Not in our library10.1063/1.4897486

5. 5.

   Schran C, Uhl F, Behler J, Marx D. Highdimensional neural network potentials for solvation: the case of protonated water clusters in helium. J. Chem. Phys. 2018;148:102310.

   Not in our library10.1063/1.4996819

6. 6.

   Behler J, Parrinello M. Generalized neural-network representation of high-dimensional potential-energy surfaces. Phys. Rev. Lett. 2007;98:146401.

   Not in our library10.1103/physrevlett.98.146401

7. 7.

   Bartók AP, Payne MC, Kondor R, Csányi G. Gaussian approximation potentials: the accuracy of quantum mechanics, without the electrons. Phys. Rev. Lett. 2010;104:136403.

   Not in our library10.1103/physrevlett.104.136403

8. 8.

   Jose KVJ, Artrith N, Behler J. Construction of high-dimensional neural network potentials using environment-dependent atom pairs. J. Chem. Phys. 2012;136:194111.

   Not in our library10.1063/1.4712397

9. 9.

   Rupp M, Tkatchenko A, Müller KR, von Lilienfeld OA. Fast and accurate modeling of molecular atomization energies with machine learning. Phys. Rev. Lett. 2012;108:058301.

   Not in our library10.1103/physrevlett.108.058301

10. 10.

    Montavon G, et al. Machine learning of molecular electronic properties in chemical compound space. New J. Phys. 2013;15:095003.

    Not in our library10.1088/1367-2630/15/9/095003

11. 11.

    Bartók AP, Kondor R, Csányi G. On representing chemical environments. Phys. Rev. B. 2013;87:184115.

    Not in our library10.1103/physrevb.87.184115

12. 12.

    Hansen K, et al. Assessment and validation of machine learning methods for predicting molecular atomization energies. J. Chem. Theory Comput. 2013;9:3404-3419.

    Not in our library10.1021/ct400195d

13. 13.

    Hansen K, et al. Machine learning predictions of molecular properties: accurate many-body potentials and nonlocality in chemical space. J. Phys. Chem. Lett. 2015;6:2326-2331.

    Not in our library10.1021/acs.jpclett.5b00831

14. 14.

    Rupp M, Ramakrishnan R, von Lilienfeld OA. Machine learning for quantum mechanical properties of atoms in molecules. J. Phys. Chem. Lett. 2015;6:3309-3313.

    Not in our library10.1021/acs.jpclett.5b01456

15. 15.

    Bartók AP, Csányi G. Gaussian approximation potentials: a brief tutorial introduction. Int. J. Quantum Chem. 2015;115:1051-1057.

    Not in our library10.1002/qua.24927

16. 16.

    Botu V, Ramprasad R. Learning scheme to predict atomic forces and accelerate materials simulations. Phys. Rev. B. 2015;92:094306.

    Not in our library10.1103/physrevb.92.094306

17. 17.

    Li Z, Kermode JR, De Vita A. Molecular dynamics with on-the-y machine learning of quantum-mechanical forces. Phys. Rev. Lett. 2015;114:096405.

    Not in our library10.1103/physrevlett.114.096405

18. 18.

    Eickenberg M, Exarchakis G, Hirn M, Mallat S, Thiry L. Solid harmonic wavelet scattering for predictions of molecule properties. J. Chem. Phys. 2018;148:241732.

    Not in our library10.1063/1.5023798

19. 19.

    Behler J. Perspective: machine learning potentials for atomistic simulations. J. Chem. Phys. 2016;145:170901.

    Not in our library10.1063/1.4966192

20. 20.

    De S, Bartok AP, Csányi G, Ceriotti M. Comparing molecules and solids across structural and alchemical space. Phys. Chem. Chem. Phys. 2016;18:13754-13769.

    Not in our library10.1039/c6cp00415f

21. 21.
    Bypassing the Kohn-Sham equations with machine learning · 2017

    No notice
22. 22.

    Artrith N, Urban A, Ceder G. Efficient and accurate machine-learning interpolation of atomic energies in compositions with many species. Phys. Rev. B. 2017;96:014112.

    Not in our library10.1103/physrevb.96.014112

23. 23.

    Podryabinkin EV, Shapeev AV. Active learning of linearly parametrized interatomic potentials. Comput. Mater. Sci. 2017;140:171-180.

    Not in our library10.1016/j.commatsci.2017.08.031

24. 24.

    Bartók AP, et al. Machine learning unifies the modeling of materials and molecules. Sci. Adv. 2017;3:e1701816.

    Not in our library10.1126/sciadv.1701816

25. 25.

    Glielmo A, Sollich P, De Vita A. Accurate interatomic force fields via machine learning with covariant kernels. Phys. Rev. B. 2017;95:214302.

    Not in our library10.1103/physrevb.95.214302

26. 26.
    Machine learning molecular dynamics for the simulation of infrared spectra · 2017

    No notice
27. 27.
    Quantum-chemical insights from deep tensor neural networks · 2017

    No notice
28. 28.

    Yao K, Herr JE, Parkhill J. The many-body expansion combined with neural networks. J. Chem. Phys. 2017;146:014106.

    Not in our library10.1063/1.4973380

29. 29.

    Dral PO, Owens A, Yurchenko SN, Thiel W. Structure-based sampling and self-correcting machine learning for accurate calculations of potential energy surfaces and vibrational levels. J. Chem. Phys. 2017;146:244108.

    Not in our library10.1063/1.4989536

30. 30.

    John S, Csányi G. Many-body coarse-grained interactions using gaussian approximation potentials. J. Phys. Chem. B. 2017;121:10934-10949.

    Not in our library10.1021/acs.jpcb.7b09636

31. 31.

    Huang, B. & von Lilienfeld, O. The “DNA” of chemistry: scalable quantum machine learning with “amons”. Preprint at https://arxiv.org/abs/1707.04146 (2017).

    Not in our library

32. 32.

    Faber FA, et al. Prediction errors of molecular machine learning models lower than hybrid DFT error. J. Chem. Theory Comput. 2017;13:5255-5264.

    Not in our library10.1021/acs.jctc.7b00577

33. 33.

    Huan TD, et al. A universal strategy for the creation of machine learning-based atomistic force fields. npj Comput. Mater. 2017;3:37.

    Not in our library

34. 34.

    Schütt K, et al. SchNet: a continuous-filter convolutional neural network for modeling quantum interactions. Adv. Neural Inf. Process. Syst. 2017;31:991-1001.

    Not in our library

35. 35.
    VAMPnets for deep learning of molecular kinetics · 2018

    Corrected
36. 36.

    Glielmo A, Zeni C, De Vita A. Efficient nonparametric n-body force fields from machine learning. Phys. Rev. B. 2018;97:184307.

    Not in our library10.1103/physrevb.97.184307

37. 37.

    Zhang L, Han J, Wang H, Car R, Weinan E. Deep potential molecular dynamics: a scalable model with the accuracy of quantum mechanics. Phys. Rev. Lett. 2018;120:143001.

    Not in our library10.1103/physrevlett.120.143001

38. 38.

    Lubbers N, Smith JS, Barros K. Hierarchical modeling of molecular energies using a deep neural network. J. Chem. Phys. 2018;148:241715.

    Not in our library10.1063/1.5011181

39. 39.

    Tang YH, Zhang D, Karniadakis GE. An atomistic fingerprint algorithm for learning ab initio molecular force fields. J. Chem. Phys. 2018;148:034101.

    Not in our library10.1063/1.5008630

40. 40.

    Grisafi A, Wilkins DM, Csányi G, Ceriotti M. Symmetry-adapted machine learning for tensorial properties of atomistic systems. Phys. Rev. Lett. 2018;120:036002.

    Not in our library10.1103/physrevlett.120.036002

41. 41.

    Ryczko K, Mills K, Luchak I, Homenick C, Tamblyn I. Convolutional neural networks for atomistic systems. Comput. Mater. Sci. 2018;149:134-142.

    Not in our library10.1016/j.commatsci.2018.03.005

42. 42.

    Kanamori K, et al. Exploring a potential energy surface by machine learning for characterizing atomic transport. Phys. Rev. B. 2018;97:125124.

    Not in our library10.1103/physrevb.97.125124

43. 43.

    Pronobis W, Tkatchenko A, Müller KR. Manybody descriptors for predicting molecular properties with machine learning: analysis of pairwise and three-body interactions in molecules. J. Chem. Theory Comput. 2018;14:2991-3003.

    Not in our library10.1021/acs.jctc.8b00110

44. 44.

    Hy TS, Trivedi S, Pan H, Anderson BM, Kondor R. Predicting molecular properties with covariant compositional networks. J. Chem. Phys. 2018;148:241745.

    Not in our library10.1063/1.5024797

45. 45.

    Smith, J. S. et al. Outsmarting quantum chemistry through transfer learning. Preprint at https://chemrxiv.org/articles/Outsmarting_Quantum_Chemistry_Through_Transfer_Learning/6744440 (2018).

    Not in our library

46. 46.

    Yao K, Herr JE, Toth DW, Mckintyre R, Parkhill J. The TensorMol-0.1 model chemistry: a neural network augmented with long-range physics. Chem. Sci. 2018;9:2261-2269.

    Not in our library10.1039/c7sc04934j

47. 47.

    Chmiela S, et al. Machine learning of accurate energyconserving molecular force fields. Sci. Adv. 2017;3:e1603015.

    Not in our library10.1126/sciadv.1603015

48. 48.

    Wilson E. B., Decius J. C., Cross P. C., Sundheim Benson R. Molecular Vibrations: The Theory of Infrared and Raman Vibrational Spectra. Journal of The Electrochemical Society. 1955;102(9):235C.

    Not in our library10.1149/1.2430134

49. 49.

    Longuet-Higgins H. The symmetry groups of non-rigid molecules. Mol. Phys. 1963;6:445-460.

    Not in our library10.1080/00268976300100501

50. 50.

    Pachauri, D., Kondor, R. & Singh, V. Solving the multi-way matching problem by permutation synchronization. Adv. Neural Inf. Process. Syst.26, 1860–1868 (2013)

    Not in our library

51. 51.

    Schiavinato M, Gasparetto A, Torsello A. Transitive Assignment Kernels for Structural Classification. 2015;146-159.

    Not in our library

52. 52.

    Kriege NM, Giscard PL, Wilson RC. On valid optimal assignment kernels and applications to graph classification. Adv. Neural Inf. Process. Syst. 2016;30:1623-1631.

    Not in our library

53. 53.

    Vishwanathan SVN, Schraudolph NN, Kondor R, Borgwardt KM. Graph kernels. J. Mach. Learn. Res. 2010;11:1201-1242.

    Not in our library

54. 54.

    Ferré G, Haut T, Barros K. Learning potential energy landscapes using graph kernels. J. Chem. Phys. 2017;146:114107.

    Not in our library10.1063/1.4978623

55. 55.

    Umeyama S. An eigendecomposition approach to weighted graph matching problems. Ieee. Trans. Pattern Anal. Mach. Intell. 1988;10:695-703.

    Not in our library10.1109/34.6778

56. 56.

    Kuhn HW. The Hungarian method for the assignment problem. Nav. Res. Logist. 1955;2:83-97.

    Not in our library10.1002/nav.3800020109

57. 57.

    González L, Mó O, Yáñez M. Density functional theory study on ethanol dimers and cyclic ethanol trimers. J. Chem. Phys. 1999;111:3855-3861.

    Not in our library10.1063/1.479689

58. 58.

    Durig J, Larsen R. Torsional vibrations and barriers to internal rotation for ethanol and 2, 2, 2-triuoroethanol. J. Mol. Struct. 1990;238:195-222.

    Not in our library10.1016/0022-2860(90)85015-b

59. 59.

    Wassermann TN, Suhm MA. Ethanol monomers and dimers revisited: a Raman study of conformational preferences and argon nanocoating effects. J. Phys. Chem. A. 2010;114:8223-8233.

    Not in our library10.1021/jp104861q

60. 60.

    Durig J, Bucy W, Wurrey C, Carreira L. Raman spectra of gases. XVI. Torsional transitions in ethanol and ethanethiol. J. Phys. Chem. A. 1975;79:988-993.

    Not in our library10.1021/j100577a009

61. 61.

    Poggio T, Anselmi F. Visual Cortex and Deep Networks: Learning Invariant Representations. 2016.

    Not in our library

62. 62.

    Anselmi F, Rosasco L, Poggio T. On invariance and selectivity in representation learning. Inf. Inference. 2016;5:134-158.

    Not in our library10.1093/imaiai/iaw009

63. 63.

    Bereau T, DiStasio RA, Tkatchenko A, Von Lilienfeld OA. Non-covalent interactions across organic and biological subsets of chemical space: physicsbased potentials parametrized from machine learning. J. Chem. Phys. 2018;148:241706.

    Not in our library10.1063/1.5009502

64. 64.

    De Luna P, Wei J, Bengio Y, Aspuru-Guzik A, Sargent E. Use machine learning to find energy materials. Nature. 2017;552:23.

    Not in our library10.1038/d41586-017-07820-6

65. 65.

    Perdew JP, Burke K, Ernzerhof M. Generalized gradient approximation made simple. Phys. Rev. Lett. 1996;77:3865-3868.

    Not in our library10.1103/physrevlett.77.3865

66. 66.

    Tkatchenko A, Scheffler M. Accurate molecular van der waals interactions from ground-state electron density and free-atom reference data. Phys. Rev. Lett. 2009;102:073005.

    Not in our library10.1103/physrevlett.102.073005

67. 67.

    Blum V, et al. Ab initio molecular simulations with numeric atom-centered orbitals. Comput. Phys. Commun. 2009;180:2175-2196.

    Not in our library10.1016/j.cpc.2009.06.022

68. 68.

    Parrish RM, et al. Psi4 1.1: an open-source electronic structure program emphasizing automation, advanced libraries, and interoperability. J. Chem. Theory Comput. 2017;13:3185-3197.

    Not in our library10.1021/acs.jctc.7b00174

69. 69.

    Ceriotti M, More J, Manolopoulos DE. i-PI: a python interface for ab initio path integral molecular dynamics simulations. Comput. Phys. Commun. 2014;185:1019-1026.

    Not in our library10.1016/j.cpc.2013.10.027

70. 70.

    Case, D. et al. Amber 2018 (The Amber Project, 2018).

    Not in our library

Cited by papers we hold 10

• Learning local equivariant representations for large-scale atomistic dynamics · 2023
  No notice
• E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials · 2022
  No notice
• Graph neural networks for materials science and chemistry · 2022
  No notice
• Gaussian Process Regression for Materials and Molecules · 2021
  No notice
• Advances in Molecular Dynamics Simulations and Enhanced Sampling Methods for the Study of Protein Systems · 2020
  No notice
• Quantum chemical accuracy from density functional approximations via machine learning · 2020
  No notice
• Approaching coupled cluster accuracy with a general-purpose neural network potential through transfer learning · 2019
  No notice
• Unifying machine learning and quantum chemistry with a deep neural network for molecular wavefunctions · 2019
  No notice
• Machine learning enables long time scale molecular photodynamics simulations · 2019
  No notice
• Unmasking Clever Hans predictors and assessing what machines really learn · 2019
  No notice

Reads

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.