arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2411.04591math.NAcs.NA
收藏

兼容有限元插值神经网络

Compatible finite element interpolated neural networks

Santiago Badia, Wei Li, Alberto F. Martín

AI总结:

本文提出兼容 FEINN,利用满足 de Rham 复形的插值空间离散 $H(\ extbf{curl})$/$H(\ extbf{div})$ PDE,并验证其在曲面问题、自适应网格及反演中的高精度与鲁棒性。

摘要:

我们将有限元插值神经网络(FEINN)框架从弱解位于 $H^1$ 的偏微分方程(PDE)扩展到弱解位于 $H(\ extbf{curl})$ 或 $H(\ extbf{div})$ 的 PDE。为此,我们考虑满足 de Rham Hilber 子复形的插值试探空间,为多种 PDE 提供稳定且保结构的神经网络离散化。该方法被命名为兼容 FEINN,已用于精确逼近 $H(\ extbf{curl})$ 内积。我们在数值上观察到,对于光滑解析解,训练后的网络比有限元解优数个数量级。此外,为展示该方法的多功能性,我们证明兼容 FEINN 在求解曲面 PDE(例如球面上的 Darcy 方程)时可达到高精度。另外,该框架能够集成自适应网格加密,以有效求解具有局部特征的问题。我们采用自适应训练策略,在一系列逐步自适应的网格上训练网络。最后,我们将兼容 FEINN 与用于求解反问题的伴随神经网络方法进行比较。我们考虑一种单循环算法,该算法使用包含 PDE 残差和数据失配项的损失函数来训练用于未知量和缺失参数的神经网络。该算法被用于根据部分或含噪观测识别 $H(\ extbf{curl})$ 模型问题的空间变化物理参数。我们发现,在这些场景中,兼容 FEINN 达到了与伴随方法相当、甚至更优的精度和鲁棒性。

英文摘要:

We extend the finite element interpolated neural network (FEINN) framework from partial differential equations (PDEs) with weak solutions in $H^1$ to PDEs with weak solutions in $H(\textbf{curl})$ or $H(\textbf{div})$. To this end, we consider interpolation trial spaces that satisfy the de Rham Hilbert subcomplex, providing stable and structure-preserving neural network discretisations for a wide variety of PDEs. This approach, coined compatible FEINNs, has been used to accurately approximate the $H(\textbf{curl})$ inner product. We numerically observe that the trained network outperforms finite element solutions by several orders of magnitude for smooth analytical solutions. Furthermore, to showcase the versatility of the method, we demonstrate that compatible FEINNs achieve high accuracy in solving surface PDEs such as the Darcy equation on a sphere. Additionally, the framework can integrate adaptive mesh refinements to effectively solve problems with localised features. We use an adaptive training strategy to train the network on a sequence of progressively adapted meshes. Finally, we compare compatible FEINNs with the adjoint neural network method for solving inverse problems. We consider a one-loop algorithm that trains the neural networks for unknowns and missing parameters using a loss function that includes PDE residual and data misfit terms. The algorithm is applied to identify space-varying physical parameters for the $H(\textbf{curl})$ model problem from partial or noisy observations. We find that compatible FEINNs achieve accuracy and robustness comparable to, if not exceeding, the adjoint method in these scenarios.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑