scientific article; zbMATH DE number 1941918
constraintsconvergenceconvexityHilbert spacenonsmoothnessoptimizationperturbationproximal projection algorithmstability
Nonsmooth analysis (49J52) Optimality conditions for problems in abstract spaces (49K27) Numerical optimization and variational techniques (65K10) Convex programming (90C25) Sensitivity, stability, parametric optimization (90C31) Approximation methods and heuristics in mathematical programming (90C59)
The paper of Y. I. Alber is a valuable and deep contribution to the algorithmical treatment of convex optimization problems in real reflexive Banach spaces and, in particular, to the corresponding stability theory. In order to quantify stability or topology among sets suitably, a number of approximation conditions are introduced and referred to throughout of the paper. In fact, there are Hausdorff-, Mosco-, super-Mosco-approximates with exponent \(p\), sub-Mosco- and over-Mosco-approximates defined with respect to a sequence of closed convex sets and to weights. In the following, these sets are feasible sets or perturbations of them. Firstly, the author analytically studies these rich and various directions of generalizing conditions and their items. Secondly, he applies these preparations to the weak convergence analysis for minimizing a convex (possibly non-smooth) functional in a Hilbert space. Here, the algorithm referred to bases on a proximal projection method with respect to stability sets perturbed. Then, provided that there is an element in the minimum point set, for the various approximation conditions and under uniform boundedness conditions on the subgradients, under estimations and summation conditions on the parameter sequences implied, conditions of the following kinds are concluded: (i) convergence of the sequence of iterate wise functional values, (ii) existence of a weak accumulation point of the sequence of iterates, (iii) all those weak accumulation points belong to the minimum point set, and (iv) the sequence of iterates itself weakly converges to an element of that set. Thirdly, a stability analysis of the strong convergence in a Hilbert space is presented in terms of Hilbert- and Mosco-approximates. This paper with its rich and well-proved results may in future serve for a deeper understanding and numerical treatment of various applications in optimization, calculus of variations and optimal control theory.
- Another version of the proximal point algorithm in a Banach space
- A proximal point method for nonsmooth convex optimization problems in Banach spaces
- Strong convergence of a proximal-based method for convex optimization
- Convergence of a Hybrid Projection-Proximal Point Algorithm Coupled with Approximation Methods in Convex Optimization
- Strong convergence and bounded perturbation resilience of a modified proximal gradient algorithm
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4408974)