Id: arxiv/2601.22897 Url: https://arxiv.org/abs/2601.22897 Tags: math.OC Score: 0 Title: Grassmannian Geometry and Global Convergence of Variable Projection for Neural Networks Domain: math.OC Source: arxiv Authors: Mathias Dus Reviews: 0 Subject: math.OC Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-01-30 Page: https://delphi.science/r/arxiv/2601.22897 Training deep neural networks and Physics-Informed Neural Networks (PINNs) often leads to ill-conditioned and stiff optimization problems. A key structural feature of these models is that they are linear in the output-layer parameters and nonlinear in the hiddenlayer parameters, yielding a separable nonlinear least-squares formulation. In this work, we study the classical variable projection (VarPro) method for such problems in the context of deep neural networks. We provide a geometric formulation on the Grassmannian and analyze the structure of critical points and convergence properties of…