Tag
Proposes KL-DNN, a scalable operator learning framework that uses Karhunen-Loève expansions to handle large-scale PDE problems, achieving lower errors and two-order-of-magnitude speedup over DeepONet on a 3D carbon storage problem.