“…Researchers have proposed some excellent approaches to predict fluid behaviors. Famous data-driven methods include regression forests [21], Fully Connected Neural Networks (FCNN) [22], [23], [24], Convolutional Neural Networks (CNN) [25], [26], [27], [28], Continuous Convolutional Neural Networks [29], [30], Recurrent Neural Networks (RNN) [31], [32], Graph Neural Networks (GNN) [33], [34], Generated Adversarial Networks (GAN) [35], [36], etc. However, little attention has arXiv:2305.03315v1 [cs.GR] 5 May 2023 been paid to tackling fluid and solid's two-way interactions since different materials' properties and interacting forces are coupled in a complicated dynamic system.…”