Functional Simulations for Uncertainty Quantification and Inverse Problems

Explore functional simulations for uncertainty quantification and inverse problems in this 48-minute conference talk from IPAM's workshop on boundary conditions for atomistic simulations in macroscopic electrochemical cells. Learn how differentiable simulations return both values and gradients with respect to input parameters, while functional simulations extend this concept to return functions valid over entire parameter space regions. Discover applications of functional atomic simulations to prototypical tasks including geometry minimization and free energy calculations, and understand how these methods enable rapid forward propagation of parameter uncertainties and novel inverse design strategies. Examine specific implementations in physical metallurgy contexts and explore how these same methodologies can be applied to atomic simulations in electrochemistry. The presentation covers recent research developments including work on uncertainty quantification methods and their practical applications in computational materials science and electrochemical systems.