Search papers, labs, and topics across Lattice.
This paper introduces a domain-decomposed randomized neural network framework specifically designed to tackle partial differential equations (PDEs) in unbounded domains, addressing the limitations of traditional truncation methods and global spectral bases. By assigning distinct subnetworks for near-field and far-field behaviors, the approach effectively captures local geometric features while managing exterior decay through a coupling mechanism based on boundary conditions. The authors validate their method through numerical experiments on Poisson and time-dependent Schr枚dinger equations, showcasing its accuracy and adaptability in complex scenarios.
A novel neural network framework achieves high accuracy in solving PDEs on unbounded domains by intelligently decomposing spatial regions into specialized subnetworks.
Partial differential equations on unbounded domains are challenging because the exterior region must be represented without excessive truncation error. Truncation-based methods often require problem-dependent artificial boundary conditions, while global spectral bases may be inefficient for localized structures, irregular geometries, or solutions with different near-field and far-field behaviors. We propose a domain-decomposed randomized neural network framework for such problems. Different randomized subnetworks are assigned to different spatial regimes: a near-field subnetwork captures local and geometric features, whereas a far-field subnetwork represents exterior decay. The subnetworks are coupled by boundary and interface conditions, and only the output-layer coefficients are solved from linear least-squares systems arising from Petrov--Galerkin or collocation formulations. We develop a Petrov--Galerkin method for semi-unbounded elliptic problems and a collocation method for fully unbounded, perforated, and time-dependent problems. A conditional bounded-parameter approximation result is proved in a broken Sobolev norm, together with an error decomposition covering approximation, empirical-consistency/quadrature, and least-squares optimization errors. Numerical experiments for Poisson and time-dependent Schr枚dinger equations demonstrate the accuracy and flexibility of the proposed method.
