Computer Guided Development of Reactions

Understanding the control features in organic reactions is critical to be able to design improved processes and to invent new processes. For the latter, understanding the canonical steps in different reaction mechanism can often be applicable outside the context originally studies. We employ both machine learning and electronic structure calculations to understand the control features in different processes. Electronic structure calculations are conducted for both homogeneous and heterogeneous catalysts. Machine learning involves both supervised learning (regression) and unsupervised learning (clustering) using Python and other programs. XSEDE supercomputing resources provided by the NSF are used to support our Gaussian, ORCA, and VASP calculations.