English

Convergence of remote projections onto convex sets

Functional Analysis 2024-01-01 v1

Abstract

Let {Cα}αΩ\{C_{\alpha}\}_{\alpha\in \Omega} be a family of closed and convex sets in a Hilbert space HH, having a nonempty intersection CC. We consider a sequence {xn}\{x_n\} of remote projections onto them. This means, x0Hx_0\in H, and xn+1x_{n+1} is the projection of xnx_n onto such a set Cα(n)C_{\alpha(n)} that the ratio of the distances from xnx_n to this set and to any other set from the family is at least tn[0,1]t_n\in [0,1]. We study properties of the weakness parameters tnt_n and of the sets CαC_\alpha which ensure the norm or weak convergence of the sequence {xn}\{x_n\} to a point in CC. We show that condition (T) is necessary and sufficient for the norm convergence of xnx_n to a point in CC for any starting element and any family of closed, convex, and symmetric sets CαC_\alpha. This generalizes a result of Temlyakov who introduced (T) in the context of greedy approximation theory. We give examples explaining to what extent the symmetry condition on the sets CαC_{\alpha} can be dropped. Condition (T) is stronger than tn2=\sum t_n^2=\infty and weaker than tn/n=\sum t_n/n=\infty. The condition tn2=\sum t_n^2=\infty turns out to be necessary and sufficient for the sequence {xn}\{x_n\} to have a partial weak limit in CC for any family of closed and convex sets CαC_\alpha and any starting element.

Keywords

Cite

@article{arxiv.2312.17574,
  title  = {Convergence of remote projections onto convex sets},
  author = {Petr A. Borodin and Eva Kopecká},
  journal= {arXiv preprint arXiv:2312.17574},
  year   = {2024}
}