Nonlinear regularity models (Q353161)

The present paper is a valuable contribution to the field of nonlinear and continuous optimization via its contribution to a qualitative understanding of such nonlinear and continuous optimization (or: programming) problems which is, then, a foundation for their quantitative, algorithmic and numerical treatment and solution. This study takes place in a wide analytic setting which allows for extended studies, including optimal control, and deep functional analytic dependences. As it relates to regularity, it means and prepares, at the same time, a twofold achievement: a generalization of the inverse and implicit function and open-mapping theorems of calculus and functional analysis, and a contribution to the theory of inverse problems of engineering. That regularity addresses equations and relations which can be interpreted in terms the feasible set of the program and of its (especially, first-order) optimality conditions. Accordingly, regularity, with its versions of different strengths in different contexts, relates closely to different constraint qualifications, (second-order) optimality conditions, to (topological, strong, structural, etc.) stability conditions of both the feasible set and the solution or stationary points, and to various concepts of complexity. Most of the results of this paper are new. The author investigates properties of regularity of set-valued mappings between metric spaces. In the context of metric regularity, nonlinear models correspond to nonlinear dependencies of estimates of error bounds in terms of residuals. Among the questions addressed, there are equivalence of the corresponding concepts of openness and ``pseudo-Hölder'' behavior, general and local regularity criteria and, for local settings, a special emphasis is on ``regularity of order \(k\)'' and, finally, in the case of so-called length range spaces, on variational methods to estimate regularity moduli. The six sections of the article are as follows: 1. Introduction, 2. Definitions and equivalences, 3. Criterion of nonlinear regularity and some applications, 4. Regularity of order \(k\), 5. Error bounds of order \(k\), and 6. A variational viewpoint. In the future, further strong insights and, additionally, methods might be expected, initiated and prepared by this research paper. Then, these achievements could support and stimulate emerging developments, particularly, in statistics and data mining, in engineering, economics, social-political decision making, finance and operation research, in medicine and healthcare.