Researchers have developed a new approach, known as physics-enhanced deep surrogate (PEDS) models, using brain-inspired neural networks to solve partial differential equations more efficiently. These equations are crucial for modeling complex physical systems with multiple rates of change, such as those involving space and time.
The traditional numerical methods for solving these equations are time-consuming and computationally intensive. The PEDS models, which integrate physics simulators to train neural networks, proved up to three times more accurate than other neural networks in solving partial differential equations. This approach reduces the required training data by at least a factor of 100, making it highly efficient for various applications in science and engineering, such as weather forecasts, carbon capture, and nuclear reactors.