arXiv 2511.05452

Self-adaptive weighting and sampling for physics-informed neural networks

By Wenqian Chen, Amanda Howard, et al.

Published 2025-11-07

Wiki summary

Explore the paper's summary, context, and related research on Papiers.

Physics-informed deep learning has emerged as a promising framework for solving partial differential equations (PDEs). Nevertheless, training these models on complex problems remains challenging, often leading to limited accuracy and efficiency. In this work, we introduce a hybrid adaptive sampling and weighting method to enhance the performance of physics-informed neural networks (PINNs). The adaptive sampling comp…

View the original paper on arXiv