Machine Learning Approach for Lowest Transition State Research of High Number Degrees of Freedom Homogeneous Catalysts

Nowadays, interesting problems in computational chemistry are found in studying complex systems with a high number of degrees of freedom. It is thus fundamental to provide a new way to handle them with a suitable tool capable of giving us the best compromise between accuracy and reliability using the minimum amount of computational time. In particular, these systems exhibit a high number of conformational isomers interconnected by low energy barriers and an accurate representation of their potential energy surface would allow us to identify the most stable isomer, the global minimum, and the transition state with the lowest energy. The challenge of this project is to provide a tool that helps us in this research, starting from a conformational analysis of four different homogeneous organocatalysts. This work aims to combine the Density Functional Theory accuracy and Molecular Mechanic Force Fields computational properties to explore the potential energy landscape using a Machine Learning approach.