Machine learning of molecular electronic properties in chemical compound space

Grégoire Montavon, Matthias Rupp, Vivekanand Gobre, Alvaro Vazquez-Mayagoitia, Katja Hansen, Alexandre Tkatchenko, Klaus Muller, O. Anatole Von Lilienfeld

Research output: Contribution to journalArticle

174 Citations (Scopus)

Abstract

The combination of modern scientific computing with electronic structure theory can lead to an unprecedented amount of data amenable to intelligent data analysis for the identification of meaningful, novel and predictive structure-property relationships. Such relationships enable high-throughput screening for relevant properties in an exponentially growing pool of virtual compounds that are synthetically accessible. Here, we present a machine learning model, trained on a database of ab initio calculation results for thousands of organic molecules, that simultaneously predicts multiple electronic ground- and excited-state properties. The properties include atomization energy, polarizability, frontier orbital eigenvalues, ionization potential, electron affinity and excitation energies. The machine learning model is based on a deep multi-task artificial neural network, exploiting the underlying correlations between various molecular properties. The input is identical to ab initio methods, i.e. nuclear charges and Cartesian coordinates of all atoms. For small organic molecules, the accuracy of such a 'quantum machine' is similar, and sometimes superior, to modern quantum-chemical methods - at negligible computational cost.

Original languageEnglish
Article number095003
JournalNew Journal of Physics
Volume15
DOIs
Publication statusPublished - 2013 Sep 1

Fingerprint

chemical compounds
machine learning
molecular electronics
Cartesian coordinates
molecular properties
atomizing
electron affinity
ionization potentials
excitation
molecules
eigenvalues
screening
electronic structure
costs
orbitals
ground state
energy
electronics
atoms

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Cite this

Montavon, G., Rupp, M., Gobre, V., Vazquez-Mayagoitia, A., Hansen, K., Tkatchenko, A., ... Anatole Von Lilienfeld, O. (2013). Machine learning of molecular electronic properties in chemical compound space. New Journal of Physics, 15, [095003]. https://doi.org/10.1088/1367-2630/15/9/095003

Machine learning of molecular electronic properties in chemical compound space. / Montavon, Grégoire; Rupp, Matthias; Gobre, Vivekanand; Vazquez-Mayagoitia, Alvaro; Hansen, Katja; Tkatchenko, Alexandre; Muller, Klaus; Anatole Von Lilienfeld, O.

In: New Journal of Physics, Vol. 15, 095003, 01.09.2013.

Research output: Contribution to journalArticle

Montavon, G, Rupp, M, Gobre, V, Vazquez-Mayagoitia, A, Hansen, K, Tkatchenko, A, Muller, K & Anatole Von Lilienfeld, O 2013, 'Machine learning of molecular electronic properties in chemical compound space', New Journal of Physics, vol. 15, 095003. https://doi.org/10.1088/1367-2630/15/9/095003
Montavon G, Rupp M, Gobre V, Vazquez-Mayagoitia A, Hansen K, Tkatchenko A et al. Machine learning of molecular electronic properties in chemical compound space. New Journal of Physics. 2013 Sep 1;15. 095003. https://doi.org/10.1088/1367-2630/15/9/095003
Montavon, Grégoire ; Rupp, Matthias ; Gobre, Vivekanand ; Vazquez-Mayagoitia, Alvaro ; Hansen, Katja ; Tkatchenko, Alexandre ; Muller, Klaus ; Anatole Von Lilienfeld, O. / Machine learning of molecular electronic properties in chemical compound space. In: New Journal of Physics. 2013 ; Vol. 15.
@article{e432b563b2f4474a96c1a4e5c12fc3d6,
title = "Machine learning of molecular electronic properties in chemical compound space",
abstract = "The combination of modern scientific computing with electronic structure theory can lead to an unprecedented amount of data amenable to intelligent data analysis for the identification of meaningful, novel and predictive structure-property relationships. Such relationships enable high-throughput screening for relevant properties in an exponentially growing pool of virtual compounds that are synthetically accessible. Here, we present a machine learning model, trained on a database of ab initio calculation results for thousands of organic molecules, that simultaneously predicts multiple electronic ground- and excited-state properties. The properties include atomization energy, polarizability, frontier orbital eigenvalues, ionization potential, electron affinity and excitation energies. The machine learning model is based on a deep multi-task artificial neural network, exploiting the underlying correlations between various molecular properties. The input is identical to ab initio methods, i.e. nuclear charges and Cartesian coordinates of all atoms. For small organic molecules, the accuracy of such a 'quantum machine' is similar, and sometimes superior, to modern quantum-chemical methods - at negligible computational cost.",
author = "Gr{\'e}goire Montavon and Matthias Rupp and Vivekanand Gobre and Alvaro Vazquez-Mayagoitia and Katja Hansen and Alexandre Tkatchenko and Klaus Muller and {Anatole Von Lilienfeld}, O.",
year = "2013",
month = "9",
day = "1",
doi = "10.1088/1367-2630/15/9/095003",
language = "English",
volume = "15",
journal = "New Journal of Physics",
issn = "1367-2630",
publisher = "IOP Publishing Ltd.",

}

TY - JOUR

T1 - Machine learning of molecular electronic properties in chemical compound space

AU - Montavon, Grégoire

AU - Rupp, Matthias

AU - Gobre, Vivekanand

AU - Vazquez-Mayagoitia, Alvaro

AU - Hansen, Katja

AU - Tkatchenko, Alexandre

AU - Muller, Klaus

AU - Anatole Von Lilienfeld, O.

PY - 2013/9/1

Y1 - 2013/9/1

N2 - The combination of modern scientific computing with electronic structure theory can lead to an unprecedented amount of data amenable to intelligent data analysis for the identification of meaningful, novel and predictive structure-property relationships. Such relationships enable high-throughput screening for relevant properties in an exponentially growing pool of virtual compounds that are synthetically accessible. Here, we present a machine learning model, trained on a database of ab initio calculation results for thousands of organic molecules, that simultaneously predicts multiple electronic ground- and excited-state properties. The properties include atomization energy, polarizability, frontier orbital eigenvalues, ionization potential, electron affinity and excitation energies. The machine learning model is based on a deep multi-task artificial neural network, exploiting the underlying correlations between various molecular properties. The input is identical to ab initio methods, i.e. nuclear charges and Cartesian coordinates of all atoms. For small organic molecules, the accuracy of such a 'quantum machine' is similar, and sometimes superior, to modern quantum-chemical methods - at negligible computational cost.

AB - The combination of modern scientific computing with electronic structure theory can lead to an unprecedented amount of data amenable to intelligent data analysis for the identification of meaningful, novel and predictive structure-property relationships. Such relationships enable high-throughput screening for relevant properties in an exponentially growing pool of virtual compounds that are synthetically accessible. Here, we present a machine learning model, trained on a database of ab initio calculation results for thousands of organic molecules, that simultaneously predicts multiple electronic ground- and excited-state properties. The properties include atomization energy, polarizability, frontier orbital eigenvalues, ionization potential, electron affinity and excitation energies. The machine learning model is based on a deep multi-task artificial neural network, exploiting the underlying correlations between various molecular properties. The input is identical to ab initio methods, i.e. nuclear charges and Cartesian coordinates of all atoms. For small organic molecules, the accuracy of such a 'quantum machine' is similar, and sometimes superior, to modern quantum-chemical methods - at negligible computational cost.

UR - http://www.scopus.com/inward/record.url?scp=84885045537&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84885045537&partnerID=8YFLogxK

U2 - 10.1088/1367-2630/15/9/095003

DO - 10.1088/1367-2630/15/9/095003

M3 - Article

VL - 15

JO - New Journal of Physics

JF - New Journal of Physics

SN - 1367-2630

M1 - 095003

ER -