Distance-Residual Physics-Informed Neural Networks: A Deep Learning Framework for Differential and Partial Differential Inclusions

License: See LICENSE file · Updated 2026-09-25
My rating

This paper introduces Distance-Residual Physics-Informed Neural Networks (DR-PINNs) for learning solutions of ordinary and partial differential inclusions. The framework constrains a differential operator to lie within an admissible set-valued map by minimizing its squared distance to that set, replacing the classical pointwise PDE or ODE residual. The resource targets scientific machine learning methods for physics-informed equation solving.

Tags: