Paperlane

no. 12 · Tuesday 10 November 2026

Convergence of optimizers implies eigenvalues filtering at equilibrium

Jerome Bolte, Quoc-Tung Le and Edouard Pauwels

math.DS · v3 · 19 pp

Abstract

Ample empirical evidence in deep neural network training suggests that a variety of optimizers tend to find nearly global optima. In this article, we adopt the reversed perspective that convergence to an arbitrary point is assumed rather than proven, focusing on the consequences of this assumption. From this viewpoint, in line with recent advances on the edge-of-stability phenomenon, we argue that different optimizers effectively act as eigenvalue filters determined by their hyperparameters. Specifically, the standard gradient descent method inherently avoids the sharpest minima, whereas Sharpness-Aware Minimization (SAM) algorithms go even further by actively favoring wider basins. Inspired by these insights, we propose two novel algorithms that exhibit enhanced eigenvalue filtering, effectively promoting wider minima. Our theoretical analysis leverages a generalized Hadamard–Perron stable manifold theorem and applies to general semialgebraic $C^2$ functions, without requiring additional non-degeneracy conditions or global Lipschitz bound assumptions. We support our conclusions with numerical experiments on feed-forward neural networks.

comments
In press, 19 pp
journal
Transactions on Machine Learning Research
fields
Dynamical Systems, Optimization and Control
pdf ↗ page arXiv:2510.09034 ↗