Daniilidis, A. (2026, April 17). Variational Stability of Alternating Projections [Presentation]. LSU Control and Optimization Zoom Seminar, Louisiana, United States of America (the).
The alternate projection method is a classical approach to dealing with the convex feasibility problem. We first show that given two nonempty closed convex sets 𝐴 and 𝐵, consecutive projections 𝑥𝑛+1 = 𝑃𝐵(𝑃𝐴(𝑥𝑛)), 𝑛 ≥ 1, produce a self-contacted sequence, providing an alternative way to establish convergence in the finite dimensional case [2]. In infinite dimensions, a regularity condition is required to ensure convergence of the sequence {𝑥𝑛} [4]. In [3], it was established that a regularity condition from [1] also ensures the variational stability of the above method. In this talk, we complete this result and show that variational stability is equivalent to the aforementioned regularity assumption.
References:
[1] H. Bauschke and J. Borwein, On the convergence of von Neumann’s alternating projection algorithm for two sets, Set-Valued Analysis 1 (1993), 185–212.
[2] A. Bohm and A. Daniilidis, Ubiquitous algorithms in convex optimization generate self-contracted sequences, Journal of Convex Analysis 29 (2022) 119–128.
[3] C. De Bernardi and E. Miglierina, A variational approach to the alternating projections method, Journal of Global Optimization 81 (2021), 323-350.
[4] H. Hundal, An alternating projection that does not converge in norm, Nonlinear Analysis: Theory, Methods, and Applications 57 (2004), 35–61.
en
Project title:
Unilateralität und Asymmetrie in der Variationsanalyse: P 36344N (FWF - Österr. Wissenschaftsfonds)