On uniqueness in structured model learning

This paper addresses the problem of uniqueness in learning physical laws for systems of partial differential equations (PDEs). Contrary to most existing approaches, it considers a framework of structured model learning, where existing, approximately correct physical models are augmented with components that are learned from data. The main results of the paper are a uniqueness and a convergence result that cover a large class of PDEs and a suitable class of neural networks used for approximating the unknown model components. The uniqueness result shows that, in the limit of full, noiseless measurements, a unique identification of the unknown model components as functions is possible as classical regularization-minimizing solutions of the PDE system. This result is complemented by a convergence result showing that model components learned as parameterized neural networks from incomplete, noisy measurements approximate the regularization-minimizing solutions of the PDE system in the limit. These results are possible under specific properties of the approximating neural networks and due to a dedicated choice of regularization. With this, a practical contribution of this analytic paper is to provide a class of model learning frameworks different to standard settings where uniqueness can be expected in the limit of full measurements.

💡 Research Summary

This paper tackles the fundamental question of whether unknown components of a physical partial differential equation (PDE) model can be uniquely identified from data when the model is only partially known. The authors introduce a “structured model learning” framework in which an existing, approximately correct physical model F(t,u,ϕ) is retained and only an additional, potentially highly nonlinear term f(t,u) is learned from measurements. This contrasts with the prevailing “full model learning” paradigm where the entire PDE is replaced by a neural network.

The first major contribution is a uniqueness theorem for the limit of perfect, noise‑free data. By formulating the identification problem as a regularized variational problem

\