This content describes a Physics-Informed Neural Network (PINN) that fuses satellite sea surface temperature (SST) with sparse in-situ loggers to resolve depth-resolved coral reef thermal fields. The model effectively corrects for overestimations of subsurface thermal stress, achieving high accuracy even with minimal training data.
Este artigo propõe uma função de perda L2 ponderada pela velocidade para resolver o modelo Bhatnagar-Gross-Krook (BGK) usando Redes Neurais Informadas pela Física (PINNs), superando as limitações da perda L2 padrão. A nova abordagem garante a convergência da solução aproximada e demonstra maior precisão e robustez em experimentos numéricos.
This paper explores solving Partial Differential Equations (PDEs) using discrete weak formulations and a discrete neural network representation. It proposes a Python environment and a DVF-CRVPINN approach for training solutions, applying discrete automatic differentiation for equations like 2D Stokes.
Physics-Informed Neural Networks (PINNs) often suffer from slow convergence and instability due to complex loss landscapes. This paper proposes a lightweight, curvature-aware optimization framework that augments existing first-order optimizers to improve convergence speed, training stability, and solution accuracy on partial differential equations (PDEs).
This paper introduces a self-supervised physics-informed neural network (PINN) framework that adaptively balances physics-based and data-driven supervision, particularly under data scarcity. It uses a learnable blending neuron to dynamically adjust term contributions based on their uncertainties and integrates transfer learning for enhanced efficiency.
This paper proposes LAM-PINN, a compositional meta-learning framework designed to mitigate task heterogeneity in Physics-Informed Neural Networks (PINNs). It addresses the challenge of training PINNs for families of partial differential equations (PDEs) which often face high computational costs or negative transfer under data-scarce conditions.