(500a) Machine Learning Reactivity in Networks of Coupled Gas Phase Reactions
AIChE Annual Meeting
2020
2020 Virtual AIChE Annual Meeting
Catalysis and Reaction Engineering Division
Reaction Engineering Perspectives: Machine Learning and Big Data for CRE (Invited Talks)
Wednesday, November 18, 2020 - 8:00am to 8:15am
The main computational cost of reaction rate constant calculations lies in the exploration of reactive pathways on potential or free energy surfaces. To overcome this challenge, recent efforts have employed machine learning to predict chemical reactivity at the level of single reactions.1â3 In this context, we have investigated the use of machine learning to predict reactivity for networks of coupled gas phase reactions.4 Single reaction rate constants were obtained by training deep neural networks on existing datasets. Chemical species were classified based on their properties and the kinetics of the reaction networks was evaluated using direct or stochastic integration of ODEs at various temperatures. This effort has enabled us to gain insight on how ML error on single rate constants influences the kinetics in a network of coupled of reactions.
- Slakman, B. L. & West, R. H. Kinetic solvent effects in organic reactions. J. Phys. Org. Chem. 32, 1â12 (2019).
- Grambow, C. A., Pattanaik, L. & Green, W. H. Deep Learning of Activation Energies. J. Phys. Chem. Lett. 11, 2992â2997 (2020).
- Komp, E. & Valleau, S. Machine learning quantum tunneling in the kinetics of chemical reactions, in preparation. (2020).
- Komp, E. & Valleau, S., in preparation. (2020).