Reduced Order
Reduced-order modeling (ROM) aims to create computationally efficient approximations of complex systems, typically governed by high-dimensional partial differential equations, by reducing the dimensionality of the problem. Current research emphasizes developing data-driven ROMs using machine learning techniques, such as autoencoders, neural networks (including recurrent and convolutional architectures), and Koopman operators, often incorporating physics-based constraints for improved accuracy and interpretability. These advancements are significantly impacting various fields by accelerating simulations, enabling real-time control of complex systems (e.g., soft robots, plasma discharges), and facilitating efficient design exploration in areas like aerospace engineering.
Papers
Latent Neural PDE Solver: a reduced-order modelling framework for partial differential equations
Zijie Li, Saurabh Patil, Francis Ogoke, Dule Shu, Wilson Zhen, Michael Schneier, John R. Buchanan,, Amir Barati Farimani
Learning reduced-order Quadratic-Linear models in Process Engineering using Operator Inference
Ion Victor Gosea, Luisa Peterson, Pawan Goyal, Jens Bremer, Kai Sundmacher, Peter Benner