It is a timely goal in the chemical sciences to computationally design novel compounds that exhibit specific properties and are straightforward to synthesize. Many relevant and promising materials properties depend explicitly on atomistic details, rendering the use of first principles simulation models mandatory. Alas, even when using high-performance computing, brute force high-throughput screening of all the possible compounds is beyond any capacity for all but the simplest systems and properties due to the combinatorial nature of chemical compound space (compositional, constitutional, and conformational isomers). In this context, structure property relationships hold great promise to dramatically assist a more rational and purposeful design of materials [1]. I will discuss recent attempts to systematically extract accurate structure property relationships for quantum chemical properties based on inductive, rather than deductive, reasoning. Given sufficient examples of properties and compounds, solutions to Schr�dinger's equation can be inferred for molecules out of sample. Specifically, I will present kernel ridge regression and neural network models that enable instantaneous predictions of atomization energies, HOMO/LUMO, polarizability, and excitation energies with an accuracy similar to hybrid density functional theory [2]. Thereafter, I will address necessary and desired requirements of molecular descriptors, as well as novel approaches, that are particularly well suited for machine learning in chemical space [3].[1] ``First principles view on chemical compound space: Gaining rigorous atomistic control of molecular properties'', O. A. von Lilienfeld, Int J Quant Chem, accepted (2012), http://arxiv.org/abs/1209.5033[2] ``Fast and accurate modeling of molecular atomization energies with machine learning'', M. Rupp, A. Tkatchenko, K.-R. M�ller, O. A. von Lilienfeld, Phys Rev Lett (2012); ``Machine Learning of Molecular Electronic Properties in Chemical Space'', G. Montavon, M. Rupp, V. Gobre, A. Vazquez-Mayagoitia, K. Hansen, A. Tkatchenko, K-R. M�ller, O. A. von Lilienfeld, PNAS, under review (2012) [3] O. A. von Lilienfeld, A. Knoll, to be submitted (2012)

"Quantum Machine'': Supervised learning of Schrödinger's equation in chemical compound space

245th National Meeting (2013)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.