Reaction: The challenge of open-shell transition metal catalysis in “systems chemistry”

Abstract

Data-driven methods have transformed the scale at which chemical transformations are being explored. This includes novel machine learning models for retrosynthesis, reaction prediction, and small-molecule generation, to name a few. Novel datasets from high-throughput computation (e.g., with first-principles density functional theory) as well as high-throughput experimentation or extraction from the literature are dramatically increasing the scale at which new compounds are discovered as well as the benefits that can be reaped from deep learning models.