Projected gradient descent accumulates at Bouligand stationary points - Archive ouverte HAL Access content directly
Preprints, Working Papers, ... Year : 2024

Projected gradient descent accumulates at Bouligand stationary points

Abstract

This paper concerns the projected gradient descent (PGD) algorithm for the problem of minimizing a continuously differentiable function on a nonempty closed subset of a Euclidean vector space. Without further assumptions, this problem is intractable and devoted algorithms are only expected to find a stationary point. PGD is known to generate a sequence whose accumulation points are Mordukhovich stationary. In this paper, these accumulation points are proven to be Bouligand stationary, and even proximally stationary if the gradient is locally Lipschitz continuous. These are the strongest stationarity properties that can be expected for the considered problem.
Fichier principal
Vignette du fichier
2403.02530v1.pdf (335.49 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04588622 , version 1 (27-05-2024)

Identifiers

Cite

Guillaume Olikier, Irène Waldspurger. Projected gradient descent accumulates at Bouligand stationary points. 2024. ⟨hal-04588622⟩
75 View
10 Download

Altmetric

Share

Gmail Mastodon Facebook X LinkedIn More