Machine learning model for non-equilibrium structures and energies of simple molecules
10.1063/1.5054968
Crossref journal-article
AIP Publishing
The Journal of Chemical Physics (317)
Abstract

Predicting molecular properties using a Machine Learning (ML) method is gaining interest among research as it offers quantum chemical accuracy at molecular mechanics speed. This prediction is performed by training an ML model using a set of reference data [mostly Density Functional Theory (DFT)] and then using it to predict properties. In this work, kernel based ML models are trained (using Bag of Bonds as well as many body tensor representation) against datasets containing non-equilibrium structures of six molecules (water, methane, ethane, propane, butane, and pentane) to predict their atomization energies and to perform a Metropolis Monte Carlo (MMC) run with simulated annealing to optimize molecular structures. The optimized structures and energies of the molecules are found to be comparable with DFT optimized structures, energies, and forces. Thus, this method offers the possibility to use a trained ML model to perform a classical simulation such as MMC without using any force field, thereby improving the accuracy of the simulation at low computational cost.

Bibliography

Iype, E., & Urolagin, S. (2019). Machine learning model for non-equilibrium structures and energies of simple molecules. The Journal of Chemical Physics, 150(2).

Authors 2
  1. E. Iype (first)
  2. S. Urolagin (additional)
References 50 Referenced 9
  1. 10.1002/qua.24927 / Int. J. Quantum Chem. / Gaussian approximation potentials: A brief tutorial introduction (2015)
  2. 10.1103/physrevb.87.184115 / Phys. Rev. B / On representing chemical environments (2013)
  3. 10.1103/physrevlett.100.185501 / Phys. Rev. Lett. / Metadynamics simulations of the high-pressure phases of silicon employing a high-dimensional neural network potential (2008)
  4. 10.1103/physrevlett.98.146401 / Phys. Rev. Lett. / Generalized neural-network representation of high-dimensional potential-energy surfaces (2007)
  5. {'issue': '1', 'key': '2023062522135990800_c5', 'first-page': '10', 'article-title': 'Simulated Annealing', 'volume': '8', 'year': '1993', 'journal-title': 'Stat. Sci.'} / Stat. Sci. / Simulated Annealing (1993)
  6. {'edition': '1st ed.', 'volume-title': 'Pattern Recognition and Machine Learning', 'year': '2006', 'key': '2023062522135990800_c6'} / Pattern Recognition and Machine Learning (2006)
  7. 10.1021/acs.jpcc.6b12752 / J. Phys. Chem. C / Modeling segregation on AuPd(111) surfaces with density functional theory and Monte Carlo simulations (2017)
  8. 10.1021/acs.jpcc.6b10908 / J. Phys. Chem. C / Machine learning force fields: Construction, validation, and outlook (2017)
  9. 10.1021/acs.jpclett.7b00774 / J. Phys. Chem. Lett. / Is the accuracy of density functional theory for atomization energies and densities in bonding regions correlated? (2017)
  10. 10.1063/1.1775767 / J. Chem. Phys. / Quantum and classical studies of vibrational motion of CH5+ on a global potential energy surface obtained from a novel ab initio direct dynamics approach (2004)
  11. 10.1039/c6cp00415f / Phys. Chem. Chem. Phys. / Comparing molecules and solids across structural and alchemical space (2016)
  12. E. Garijo del Río, J. Jørgen Mortensen, and K. W. Jacobsen, “A local Bayesian optimizer for atomic structures,” e-print arXiv:1808.08588 (2018). (10.1103/PhysRevB.100.104103)
  13. {'volume-title': 'Pattern Classification', 'year': '2001', 'key': '2023062522135990800_c13'} / Pattern Classification (2001)
  14. 10.1063/1.5020710 / J. Chem. Phys. / Alchemical and structural distribution based representation for universal quantum machine learning (2018)
  15. 10.1103/physrevlett.117.135502 / Phys. Rev. Lett. / Machine learning energies of 2 million elpasolite (ABC2D6) crystals (2016)
  16. 10.1007/s002140050021 / Theor. Chem. Acc. / Towards an order-N DFT method (1998)
  17. 10.1021/ct800468h / J. Chem. Theory Comput. / Dynamically polarizable water potential based on multipole moments trained by machine learning (2009)
  18. 10.1021/acs.jpclett.5b00831 / J. Phys. Chem. Lett. / Machine learning predictions of molecular properties: Accurate many-body potentials and nonlocality in chemical space (2015)
  19. 10.1021/ct400195d / J. Chem. Theory Comput. / Assessment and validation of machine learning methods for predicting molecular atomization energies (2013)
  20. 10.1021/cm100795d / Chem. Mater. / Finding natures missing ternary oxide compounds using machine learning and density functional theory (2010)
  21. 10.1063/1.4964627 / J. Chem. Phys. / Communication: Understanding molecular representations in machine learning: The role of uniqueness and target similarity (2016)
  22. H. Huo and M. Rupp, “Unified representation of molecules and crystals for machine learning,” e-print arXiv:1704.06439 (2017).
  23. 10.1002/jcc.23246 / J. Comput. Chem. / Parameterization of a reactive force field using a Monte Carlo algorithm (2013)
  24. 10.1021/acs.jpclett.7b02010 / J. Phys. Chem. Lett. / Predicting catalytic activity of nanoparticles by a DFT-aided machine-learning algorithm (2017)
  25. 10.1021/acs.jpcc.6b05068 / J. Phys. Chem. C / Machine learning assisted predictions of intrinsic dielectric breakdown strength of ABX3 perovskites (2016)
  26. 10.1021/acs.jpca.7b01182 / J. Phys. Chem. A / Representing global reactive potential energy surfaces using Gaussian processes (2017)
  27. 10.1016/s0166-1280(96)04810-5 / J. Mol. Struct.: THEOCHEM / Simulated annealing study of cation distribution in dehydrated zeolites (1997)
  28. 10.1021/jp0221993 / J. Phys. Chem. A / Robust and affordable multicoefficient methods for thermochemistry and thermochemical kinetics: The MCCM/3 suite and SAC/3 (2003)
  29. {'edition': '7th ed.', 'volume-title': 'Information Theory, Inference, and Learning Algorithms', 'year': '2005', 'key': '2023062522135990800_c29'} / Information Theory, Inference, and Learning Algorithms (2005)
  30. 10.1063/1.1699114 / J. Chem. Phys. / Equation of state calculations by fast computing machines (1953)
  31. {'volume-title': 'Machine Learning', 'year': '1997', 'key': '2023062522135990800_c31'} / Machine Learning (1997)
  32. 10.1088/1367-2630/15/9/095003 / New J. Phys. / Machine learning of molecular electronic properties in chemical compound space (2013)
  33. 10.1103/physrevb.45.13244 / Phys. Rev. B / Accurate and simple analytical representation of the electron-gas correlation energy (1992)
  34. 10.1021/acs.jctc.5b00099 / J. Chem. Theory Comput. / Big data meets quantum chemistry approximations: The Δ-machine learning approach (2015)
  35. 10.1063/1.4928757 / J. Chem. Phys. / Electronic spectra from TDDFT and machine learning in chemical space (2015)
  36. 10.1007/978-0-387-39940-9_565 / Encyclopedia of Database systems / Cross-validation (2009)
  37. 10.1002/qua.24954 / Int. J. Quantum Chem. / Machine learning for quantum mechanics in a nutshell (2015)
  38. 10.1021/acs.jpclett.5b01456 / J. Phys. Chem. Lett. / Machine learning for quantum mechanical properties of atoms in molecules (2015)
  39. 10.1103/physrevlett.108.058301 / Phys. Rev. Lett. / Fast and accurate modeling of molecular atomization energies with machine learning (2012)
  40. 10.1063/1.5043213 / J. Chem. Phys. / Guest editorial: Special topic on data-enabled theoretical chemistry (2018)
  41. {'key': '2023062522135990800_c41', 'first-page': '301', 'article-title': 'The kernel trick for distances', 'author': 'Leen', 'year': '2001', 'journal-title': 'Advances in Neural Information Processing Systems 13'} / Advances in Neural Information Processing Systems 13 / The kernel trick for distances by Leen (2001)
  42. 10.1137/15m1054183 / Multiscale Model. Simul. / Moment tensor potentials: A class of systematically improvable interatomic potentials (2015)
  43. 10.1002/jcc.1056 / J. Comput. Chem. / Chemistry with ADF (2001)
  44. 10.1002/jcc.10255 / J. Comput. Chem. / Optimized Slater-type basis sets for the elements 1-118 (2003)
  45. 10.1021/ci200409x / J. Chem. Inf. Model. / Machine learning methods for property prediction in chemoinformatics: Quo vadis? (2012)
  46. 10.1002/qua.24375 / Int. J. Quantum Chem. / First principles view on chemical compound space: Gaining rigorous atomistic control of molecular properties (2013)
  47. 10.1002/qua.24912 / Int. J. Quantum Chem. / Fourier series of atomic radial distribution functions: A molecular fingerprint for machine learning models of quantum chemical properties (2015)
  48. 10.1002/qua.24939 / Int. J. Quantum Chem. / Understanding kernel ridge regression: Common behaviors from simple functions to density functionals (2015)
  49. 10.1002/bip.360290127 / Biopolymers / Applications of simulated annealing to peptides (1990)
  50. 10.1021/acs.jpclett.7b01072 / J. Phys. Chem. Lett. / Intrinsic bond energies from a bonds-in-molecules neural network (2017)
Dates
Type When
Created 6 years, 7 months ago (Jan. 10, 2019, 1:33 p.m.)
Deposited 2 years, 1 month ago (June 25, 2023, 6:14 p.m.)
Indexed 3 weeks, 1 day ago (July 30, 2025, 7:10 a.m.)
Issued 6 years, 7 months ago (Jan. 10, 2019)
Published 6 years, 7 months ago (Jan. 10, 2019)
Published Online 6 years, 7 months ago (Jan. 10, 2019)
Published Print 6 years, 7 months ago (Jan. 14, 2019)
Funders 0

None

@article{Iype_2019, title={Machine learning model for non-equilibrium structures and energies of simple molecules}, volume={150}, ISSN={1089-7690}, url={http://dx.doi.org/10.1063/1.5054968}, DOI={10.1063/1.5054968}, number={2}, journal={The Journal of Chemical Physics}, publisher={AIP Publishing}, author={Iype, E. and Urolagin, S.}, year={2019}, month=jan }