Quantum chemistry and machine learning in computational materials and interface chemistry

Abstract

Due to the ever-increasing improvement in computer power and the development of more efficient codes, atomistic first-principles calculations have become an indispensable tool in materials and interface chemistry. They are no longer limited to explanatory purposes but have gained predictive power, so that computational modeling and experiment can collaborate on the same footing. Still, quantum chemical first-principles methods are computationally expensive, and there are certain limitations as far as their scaling behavior in high-performance computing is concerned. Hence computational methods based on machine learning have become increasingly popular as an alternative approach to study materials and interfaces. Here some success stories of both approaches will be presented and their respective advantages and disadvantages will be critically discussed.