boundary-conditions-for-pinos-code

We introduce a way to prescribe Neumann or Robin boundary conditions when solving partial differential equations with physics-informed machine learning on a domain with an only piecewise differentiable boundary that avoids the stability issues of existing approaches.我们提出了一种方法，在基于物理信息机器学习（Physics-Informed Machine Learning）求解边界仅为分段可微的区域上的偏微分方程时，可指定诺伊曼（Neumann）边界条件或罗宾（Robin）边界条件，该方法能够规避现有方法所存在的稳定性问题。