On Benchmarking of Automated Methods for Performing Exhaustive Reaction Path Search
journal contributionposted on 12.03.2019, 00:00 by Satoshi Maeda, Yu Harabuchi
In recent years, the importance of computational chemistry approaches has grown rapidly because of recent advances in computational software and hardware. Automated reaction path search is one of the promising techniques which would allow exploration of unknown chemistry using computers. Several methods have been developed so far, and comparison of the performance of existing methods would be a subject of interest to many chemists. In this paper, we present the performance of our single-component artificial force induced reaction method (SC-AFIR) to rectify the result shown in a recent report [Grambow et al. J. Am. Chem. Soc. 2018, 140, 1035−1048] on the performance of the SC-AFIR method compared to four other methods. We discuss the flaws in their benchmark procedure and our thoughts on benchmarking of automated reaction path search methods.