- Physical Review Journals
All JournalsPhysics Magazine

Physical Review B

Reuse & Permissions

It is not necessary to obtain permission to reuse this article or its components as it is available under the terms of the Creative Commons Attribution 4.0 International license. This license permits unrestricted use, distribution, and reproduction in any medium, provided attribution to the author(s) and the published article's title, journal citation, and DOI are maintained. Please note that some figures may have been included with permission from other third parties. It is your responsibility to obtain the proper permission from the rights holder directly for these figures.

Export citation

Export citation

Choose format for download:

Download Citation
  • Letter
  • Open Access

High-pressure hydrogen by machine learning and quantum Monte Carlo

Andrea Tirelli1,*, Giacomo Tenti1, Kousuke Nakano1,2, and Sandro Sorella1,3,†

  • 1International School for Advanced Studies (SISSA), Via Bonomea 265, 34136 Trieste, Italy
  • 2School of Information Science, JAIST, Asahidai 1-1, Nomi, Ishikawa 923-1292, Japan
  • 3Computational Materials Science Research Team, RIKEN Center for Computational Science (R-CCS), Kobe, Hyogo 650-0047, Japan

  • *atirelli@sissa.it
  • †sorella@sissa.it
PDF

Phys. Rev. B 106, L041105 – Published 8 July, 2022

DOI: https://doi.org/10.1103/PhysRevB.106.L041105

Abstract

We have developed a technique combining the accuracy of quantum Monte Carlo in describing the electron correlation with the efficiency of a machine learning potential (MLP). We use kernel regression in combination with the smooth overlap of atomic position (SOAP) features, implemented here in a very efficient way. The key ingredients are as follows: (i) a sparsification technique, based on farthest point sampling, ensuring generality and transferability of our MLPs, and (ii) the so-called Δ–learning, allowing a small training data set, a fundamental property for highly accurate but computationally demanding calculations, such as the ones based on quantum Monte Carlo. As an application we present a benchmark study of the liquid-liquid transition of high-pressure hydrogen and show the quality of our MLP, by emphasizing the importance of high accuracy for this very debated subject, where experiments are difficult in the laboratory, and theory is still far from being conclusive.

View figure in article

Physics Subject Headings (PhySH)

Article Text

Supplemental Material

References (43)

  1. E. Wigner and H. Huntington, J. Chem. Phys. 3, 764 (1935).
  2. A. Drozdov, M. Eremets, I. Troyan, V. Ksenofontov, and S. I. Shylin, Nature (London) 525, 73 (2015).
  3. M. Somayazulu, M. Ahart, A. K. Mishra, Z. M. Geballe, M. Baldini, Y. Meng, V. V. Struzhkin, and R. J. Hemley, Phys. Rev. Lett. 122, 027001 (2019).
  4. N. D. Drummond, B. Monserrat, J. H. Lloyd-Williams, P. L. Ríos, C. J. Pickard, and R. J. Needs, Nat. Commun. 6, 7794 (2015).
  5. A. Drozdov, P. Kong, V. Minkov, S. Besedin, M. Kuzovnikov, S. Mozaffari, L. Balicas, F. Balakirev, D. Graf, V. Prakapenka et al., Nature (London) 569, 528 (2019).
  6. P. M. Celliers, M. Millot, S. Brygoo, R. S. McWilliams, D. E. Fratanduono, J. R. Rygg, A. F. Goncharov, P. Loubeyre, J. H. Eggert, J. L. Peterson et al., Science 361, 677 (2018).
  7. M. D. Knudson, M. P. Desjarlais, A. Becker, R. W. Lemke, K. Cochrane, M. E. Savage, D. E. Bliss, T. Mattsson, and R. Redmer, Science 348, 1455 (2015).
  8. K. Ohta, K. Ichimaru, M. Einaga, S. Kawaguchi, K. Shimizu, T. Matsuoka, N. Hirao, and Y. Ohishi, Sci. Rep. 5, 16560 (2015).
  9. R. S. McWilliams, D. A. Dalton, M. F. Mahmood, and A. F. Goncharov, Phys. Rev. Lett. 116, 255501 (2016).
  10. M. Zaghoo, A. Salamat, and I. F. Silvera, Phys. Rev. B 93, 155128 (2016).
  11. M. Zaghoo and I. F. Silvera, Proc. Natl. Acad. Sci. USA 114, 11873 (2017).
  12. S. Scandolo, Proc. Natl. Acad. Sci. USA 100, 3051 (2003).
  13. M. A. Morales, C. Pierleoni, E. Schwegler, and D. M. Ceperley, Proc. Natl. Acad. Sci. USA 107, 12799 (2010).
  14. W. Lorenzen, B. Holst, and R. Redmer, Phys. Rev. B 82, 195107 (2010).
  15. K. T. Delaney, C. Pierleoni, and D. M. Ceperley, Phys. Rev. Lett. 97, 235702 (2006).
  16. J. Vorberger, I. Tamblyn, B. Militzer, and S. A. Bonev, Phys. Rev. B 75, 024206 (2007).
  17. G. Mazzola, R. Helled, and S. Sorella, Phys. Rev. Lett. 120, 025701 (2018).
  18. H. Y. Geng, Q. Wu, M. Marqués, and G. J. Ackland, Phys. Rev. B 100, 134109 (2019).
  19. B. Cheng, G. Mazzola, C. J. Pickard, and M. Ceriotti, Nature (London) 585, 217 (2020).
  20. W. Jia, H. Wang, M. Chen, D. Lu, L. Lin, R. Car, W. E, and L. Zhang, Pushing the limit of molecular dynamics with ab initio accuracy to 100 million atoms with machine learning, in Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis (IEEE, New York, 2020).
  21. J. Behler and M. Parrinello, Phys. Rev. Lett. 98, 146401 (2007).
  22. J. Behler, J. Chem. Phys. 134, 074106 (2011).
  23. W. Foulkes, L. Mitas, R. Needs, and G. Rajagopal, Rev. Mod. Phys. 73, 33 (2001).
  24. S. Azadi and W. M. C. Foulkes, Phys. Rev. B 88, 014115 (2013).
  25. R. C. Clay, J. Mcminis, J. M. McMahon, C. Pierleoni, D. M. Ceperley, and M. A. Morales, Phys. Rev. B 89, 184106 (2014).
  26. A. P. Bartók, M. C. Payne, R. Kondor, and G. Csányi, Phys. Rev. Lett. 104, 136403 (2010).
  27. See Supplemental Material at http://link.aps.org/supplemental/10.1103/PhysRevB.106.L041105 for technical details of the methods, benchmark, and validation tests.
  28. A. P. Bartók, R. Kondor, and G. Csányi, Phys. Rev. B 87, 184115 (2013).
  29. f(rij)=1 if 0<rij≤rc/2 and f(rij)=4(rij/rc−1)2(4rij/rc−1) for rc/2≤rij≤rc. We also assume that f(0)=0, in order to exclude the trivial i=j contribution.
  30. L. Mitáš, E. L. Shirley, and D. M. Ceperley, J. Chem. Phys. 95, 3467 (1991).
  31. J. Kleinberg and E. Tardos, Algorithm Design (Pearson Education India, Delhi, 2006).
  32. R. Ramakrishnan, P. O. Dral, M. Rupp, and O. A. von Lilienfeld, J. Chem. Theory Comput. 11, 2087 (2015).
  33. P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli, G. L. Chiarotti, M. Cococcioni, I. Dabo et al., J. Phys.: Condens. Matter 21, 395502 (2009).
  34. K. Nakano, C. Attaccalite, M. Barborini, L. Capriotti, M. Casula, E. Coccia, M. Dagrada, C. Genovese, Y. Luo, G. Mazzola, A. Zen, and S. Sorella, J. Chem. Phys. 152, 204121 (2020).
  35. This error bar can be achieved, for instance, by a single QMC simulation on about 1000 processors for about a couple of hours within the turborvb package [34]. Clearly this error bar cannot be sustained for an MD run with several thousand steps, as the ones employed here.
  36. N. V. Kampen, in Stochastic Processes in Physics and Chemistry, edited by N. Van Kampen, North-Holland Personal Library, 3rd ed. (Elsevier, Amsterdam, 2007), pp. 219–243.
  37. G. Bussi, D. Donadio, and M. Parrinello, J. Chem. Phys. 126, 014101 (2007).
  38. C. Pierleoni, M. A. Morales, G. Rillo, M. Holzmann, and D. M. Ceperley, Proc. Natl. Acad. Sci. USA 113, 4953 (2016).
  39. V. V. Karasiev, J. Hinz, S. X. Hu, and S. B. Trickey, Nature (London) 600, E12 (2021).
  40. J. Hinz, V. V. Karasiev, S. X. Hu, M. Zaghoo, D. Mejía-Rodríguez, S. B. Trickey, and L. Calderín, Phys. Rev. Research 2, 032065(R) (2020).
  41. A. Rohatgi, Webplotdigitizer: Version 4.5, https://automeris.io/WebPlotDigitizer (2021).
  42. K. Nakano, A. Raghav, and S. Sorella, J. Chem. Phys. 156, 034101 (2022).
  43. S. R. Xie, M. Rupp, and R. G. Hennig, arXiv:2110.00624.

Outline

Information

Phys. Rev. B 106, L041105– Published 8 July, 2022
  • Received 22 December 2021
  • Accepted 13 June 2022
Crossmark - Check for updates button
Creative Commons Logo

Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.

©2022 American Physical Society

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation