Comparing Physics-informed Neural Networks and Discrete Singular Convolution for Unsteady Open-channel Flow
Year: 2026
Type: article
Abstract: Supplemental Materials (Sections S1–S10, Eqs. S1–S19, Tables S1–S7, Figs. S1–S2) for the paper "Comparing Physics-Informed Neural Networks and Discrete Singular Convolution for Unsteady Open-Channel Flow" by G. Gurarslan, E. Bahar, and A. Kurt, submitted to the Journal of Hydraulic Engineering (ASCE). Contents: construction of the discrete singular convolution operator; characteristic boundary rows and their Jacobian; formulations of the finite element, space–time differential quadrature, and Preissmann comparison methods; implementation, network architecture, and training procedure of the physics-informed neural network; Richardson extrapolation of the reference solutions; additional verification (parameter sensitivity, linearized stability of mapped grids, manufactured solution, global mass balance, time discretization of the comparison methods); convergence and computational-cost tables; notation.
Authors: Gurarslan, Gurhan, Bahar, Ersin, Kurt, Ali
Institution: Pamukkale University
Language: English
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225450835