The development of a new methodology, known as Accurate NeurAl networK engINe for Molecular Energies (ANAKIN-ME, or ANI for short), is able, it is claimed, to describe the forces in molecules as accurately as density functional theory (DFT), but hundreds of thousands of times faster. This combination of speed and accuracy could allow researchers to tackle problems that were previously impossible, leading to breakthroughs in the arenas of drug discovery and materials science. Details of the method by J. S. Smith O. Isayev and A. E. Roitberg built on earlier work of Michele Parrinello are available in a 2017 publication entitled “ANI-1: an extensible neural network potential with DFT accuracy at force field computational cost” arXiv:1610.08935v4 .
The methodology has been ported to NVIDIA, and will be the subject of a webinar hosted by NVIDIA, University of Florida, University of North Carolina, on 20 September 2017 from 10am-11am PST . Sign up for the webinar here.