S.D. Edkins
Citrine Informatics,
United States
Keywords: machine learning, artificial intelligence, chemicals, materials, modeling, simulation
Summary:
Physics-based models (such as density functional theory, molecular dynamics, and computational fluid dynamics) are used to simulate and study the behavior of materials and chemical systems at the mechanistic level based on theory. Conversely, empirical machine learning models rely on statistical correlations, which allows them to be predictive even in the absence of a useful or tractable theoretical description. Combining these techniques, we get the best of both worlds. In this talk I will use case studies to illustrate how Citrine’s customers have combined simulation and machine learning to accelerate product development, achieving results like decreasing the time to remove PFAS from products by 40%.