Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces Kolmogorov-Arnold Networks (KANs) as a novel approach for solving free-boundary partial differential equations (PDEs) within a physics-informed framework. By integrating obstacle constraints, PDE inequalities, complementarity conditions, and boundary conditions into residual-based loss functions, KANs are applied to linear elliptic, nonlinear $p$-Laplacian obstacle problems, and a Stefan problem. Numerical results demonstrate that KANs achieve superior accuracy in terms of $L^2$ and $L^\infty$ errors and accurately capture complex phenomena like contact regions and moving interfaces, outperforming existing PINN and residual-network baselines.
Free-boundary PDEs are solved with unprecedented accuracy and interface resolution using Kolmogorov-Arnold Networks, surpassing PINNs and residual networks.
We study free-boundary problems within a physics-informed framework using Kolmogorov-Arnold network (KAN) approximations. The proposed approach incorporates obstacle constraints, partial differential equation (PDE) inequalities, complementarity conditions, and boundary conditions through residual-based loss functions. We consider a linear elliptic obstacle problem, a nonlinear $p$-Laplacian obstacle problem, and a time-dependent one-phase Stefan problem. The proposed KAN solver is compared with physics-informed neural network (PINN) and residual-network baselines. Numerical experiments show that KANs achieve low relative $L^2$ and $L^\infty$ errors while accurately resolving contact regions and moving interfaces. The results indicate that KAN representations provide an effective alternative for solving free-boundary PDEs.