Machine learning of molecular electronic properties in chemical compound space
-
Montavon, Gregoire[Machine Learning Group, Technical University of Berlin, Marchstraße 23, D-10587 Berlin, Germany]
Rupp, Matthias[Institute of Pharmaceutical Sciences, ETH Zurich, CH 8093 Z ̈ urich, Switzerland]
Gobre, Vivekanand[Fritz-Haber-Institut der Max-Planck-Gesellschaft, D-14195 Berlin, Germany]
Vazquez-Mayagoitia, Alvaro[Argonne Leadership Computing Facility, Argonne National Laboratory, Argonne, IL 0439, USA]
Hansen, Katja[Fritz-Haber-Institut der Max-Planck-Gesellschaft, D-14195 Berlin, Germany]
Tkatchenko, Alexandre[Fritz-Haber-Institut der Max-Planck-Gesellschaft, D-14195 Berlin, Germany > > > ; Department of Chemistry, Pohang University of Science and Technology, Pohang 790-784, Korea]
Mueller, Klaus-Robert[Machine Learning Group, Technical University of Berlin, Marchstraße 23, D-10587 Berlin, Germany > > > ; Department of Brain and Cognitive Engineering, Korea University, Anam-dong, Seongbuk-gu, Seoul 136-713, Korea]
von Lilienfeld, O. Anatole[Argonne Leadership Computing Facility, Argonne National Laboratory, Argonne, IL 0439, USA]
TEMPLE CIRCUS, TEMPLE WAY, BRISTOL BS1 6BE, ENGLAND
[en] 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.