Learning universal approximations for partial differential equations with Physics-Informed Broad Learning System
Researchers have introduced the Physics-Informed Broad Learning System (PIBLS), a novel framework designed to solve partial differential equations (PDEs) more efficiently than existing methods. Unlike traditional numerical solvers that can be computationally expensive or Physics-Informed Neural Networks (PINNs) that may suffer from slow convergence, PIBLS reformulates PDE solving as a direct least-squares optimization. This backpropagation-free approach has demonstrated the ability to be one to three orders of magnitude faster than conventional PINNs while achieving higher accuracy, offering a practical alternative for real-time scientific simulations. AI
IMPACT This framework offers a computationally efficient paradigm for scientific machine learning, potentially accelerating real-time simulation and design optimization tasks.