Abstract by Samuel Pugh
Chemistry and Biochemistry
Using Machine Learning to Predict Dynamical Product Ratios from Quasiclassical Direct Dynamics Simulations
This talk will outline the use of machine learning algorithms to predict selectivity in the diazotization of 2,3-diazabicyclo[2.2.1]hept-2-ene. The use of our in house quasiclassical direct dynamics program DynSuite to model dynamic trajectories and record vibrational energies and atomic velocities will be discussed. Finally, feature selection, and the efficacy of various machine learning algorithms to predict product ratios for this reaction will be reported.