2026-10-06T19:24:24Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34046457 2026-10-01T14:59:16Z category_28813 category_29164 item_type_1 month_year_10_2026

Manufactured problem design for unbounded inverse problems Gregorio Perez Bernal (22258905) Oscar Rincón Cardeño (22258780) Silvana Montoya-Noguera (22258901) Nicolas Guarin-Zapata (6532391) Applications in physical sciences Neural networks Scientific machine learning (SciML) Physics-Informed Neural Networks (PINNs) Kolmogorov-Arnold Networks(KAN) Inverse problems (Differential equations) Unbounded Domains (UDs) <p dir="ltr">(Top) Schematic workflow illustrating the formulation process, progressing from target definition and domain specification (infinite and semi-infinite) to the generation of a manufactured solution satisfying the Poisson equation −∇ · (k ∇u) = f. (Bottom) Analytical solutions u and spatially varying coefficient fields k. The infinite-domain configuration (left) uses parameters α = 0.5, β = 5, and ε = 1, whereas the semi-infinite-domain configuration (right) uses α = 0.5, β = 5, and ε = 0.75.</p> 2026-10-01T14:59:16Z Image Figure 10.6084/m9.figshare.34046457.v1 https://figshare.com/articles/figure/Manufactured_problem_design_for_unbounded_inverse_problems/34046457 CC BY 4.0