1: \begin{abstract}

2: 

3: Modeling dynamics in the form of partial differential equations (PDEs) is an effectual way to understand real-world physics processes.

4: For complex physics systems, analytical solutions are not available and numerical solutions are widely-used.

5: However, traditional numerical algorithms are computationally expensive and challenging in handling multiphysics systems.

6: Recently, using neural networks to solve PDEs has made significant progress, called physics-informed neural networks (PINNs).

7: PINNs encode physical laws into neural networks and learn the continuous solutions of PDEs.

8: For the training of PINNs, existing methods suffer from the problems of inefficiency and unstable convergence, since the PDE residuals require calculating automatic differentiation.

9: In this paper, we propose \textbf{D}ynamic \textbf{M}esh-based \textbf{I}mportance \textbf{S}ampling (DMIS) to tackle these problems.

10: DMIS is a novel sampling scheme based on importance sampling, which constructs a dynamic triangular mesh to estimate sample weights efficiently.

11: DMIS has broad applicability and can be easily integrated into existing methods.

12: The evaluation of DMIS on three widely-used benchmarks shows that DMIS improves the convergence speed and accuracy in the meantime.

13: Especially in solving the highly nonlinear Schrödinger Equation, compared with state-of-the-art methods, DMIS shows up to 46\% smaller root mean square error and five times faster convergence speed.

14: Code is available at https://github.com/MatrixBrain/DMIS.

15: 

16: \end{abstract}

17: