Set-valued evenly convex functions: characterizations and c-conjugacy

M.D. Fajardo · 2025

Read Set-valued evenly convex functions: characterizations and c-conjugacy by M.D. Fajardo (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-06079.txt.

Introduction

M D Fajardo studies functions that return sets rather than single values, in a 2025 arXiv paper. Such functions arise where a problem has ranges of answers ranked by a cone's order. She focuses on evenly convex ones, whose graph's upper region can be cut out by open half-spaces, a shape notion weaker than plain convexity.

Fajardo proves that each such function is exactly the pointwise top of its flat minorants, the set-valued affine functions that sit under it. She then builds a matching scheme for this class, pairing each function with a dual built from those minorants. The scheme obeys a double-dual law in the spirit of the classic Fenchel-Moreau theorem: applying it twice returns the function itself. The work is pure theory in convex analysis.

Catalog record

CREATORM.D. Fajardo
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE54 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-06079

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Set-valued evenly convex functions: characterizations and c-conjugacy M.D. Fajardo Abstract In this work we deal with set-valued functions with values in the power set of a separated locally convex space where a nontrivial pointed convex cone induces a partial order relation. A set-valued function is evenly convex if its epigraph is an evenly convex set, i.e., it is the intersection of an arbitrary family of open half-spaces. In this paper we characterize evenly convex set-valued functions as the pointwise supremum of its set-valued e-affine minorants. Moreover, a suitable conjugation pattern will be developed for these functions, as well as the counterpart of the biconjugation Fenchel-Moreau theorem. ∎ MSC 49N15 52A41 90C25 1 Introduction Evenly convex sets (e-convex in brief) were introduced by Fenchel F1952 in \mathbb{R}^{n} as those sets which can be expressed as the intersection of an arbitrary (possibly empty) family of open half-spaces. He initiated the study of this kind of sets defining a natural polarity operation and mentioning some of their properties. Later, they appeared in some works related to quasiconvex programming (see Ml1983 , Ml1988 , PP1984 and PV1990 ) and mathematical economics ML1991 . E-convex sets are the solution sets of linear systems of either weak or strict inequalities, and from this point of view, their properties have been studied in GJR2003 and GR2006 , whereas some important characterizations of e-convex sets in terms of their sections and projections were given in KMZ2007 . Finally, some properties of this class of sets in locally convex spaces can be found in DML2002 . In a natural way, the concept of even convexity was applied to an extended real-valued function defined on \mathbb{R}^{n} in RVP2011 , where a function is said to be e-convex if its epigraph is e-convex. In that work, the main properties of these functions, which extend the class of convex and lower semicontinuous functions, were studied. In particular, some characterizations were provided. Later, in MLVP2011 , the notion of e-convex function was considered for extended real-valued functions defined on locally convex spaces, and proper e-convex functions were characterized as the pointwise supremum of their e-affine minorants. This characterization allowed, also in MLVP2011 , the definition of a suitable conjugation scheme for e-convex functions. This pattern is called c -conjugation , and it was inspired by a survey written by Martínez-Legaz ML2005 where generalized convex duality theory is applied to quasiconvex programming. As it is shown in MLVP2011 , the c -conjugation scheme allowed to set forth the e-convex counterpart of the well-known Fenchel-Moreau theorem. Recently, some works have been developed around duality theory in evenly convex optimization ( F2015 , FVR2016 , FV2016SSD , FV2017 , FV2018 ), building different c-conjugate dual problems by means of the perturbation approach and obtaining regularity conditions expressed in terms of the even convexity of the perturbation function. A monograph presenting the state-of-art on even convexity in finite dimensional spaces has been published very recently (see FGRV2020 ). In this paper we extend the notion of even convexity to set-valued functions which map a separated locally convex space X into the power set \mathcal{P}(Z) of another separated locally convex space Z , assuming that there exists a partial order in Z induced by a pointed convex cone K\subset Z . Every concept and result from the even convexity scalar theory can be translated in the field of set-valued functions by choosing appropiate definitions of properness, e-affine set-valued minorants, set-valued c-conjugate function, etc, and, moreover, extending operations, like the meaning of pointwise supremum or infimum. We will assume the same framework as in H2009 , where a characterization for proper closed convex set-valued functions by its continuous affine minorants and an appropiate Fenchel conjugate are provided. Also in that paper introduction section, different approaches which have been developed for conjugation patterns suitable for closed convex set-valued functions are highlighted. The layout of this work is as follows. Section 2 contains preliminary results on set-valued functions, as well as fundamental definitions and results on e-convex sets and extended real-valued functions to make the paper self-contained. Section 3 is dedicated to characterizations of set-valued proper and e-convex functions in terms of different kinds of affine minorants. Section 4 is devoted to a suitable conjugation pattern for set-valued e-convex functions, which will provide a biconjugation theorem. 2 Preliminaries Let X and Z be nontrivial separated locally convex spaces, we denote by X^{*} and Z^{*} their respective topological dual spaces, and by \langle\cdot,\cdot\rangle the duality product in both cases: \langle x,x^{*}\rangle=x^{*}(x) for all x\in X,x^{*}\in X^{*} (analogously in Z and Z^{*} ). In this paper we will deal with set-valued functions f:X\to\mathcal{P}(Z) , where \mathcal{P}(Z) is the set of all the subsets of Z , including the empty set. The algebraic structure of \mathcal{P}(Z) is determined by the Minkowski sum A+B=\{a+b:a\in A,b\in B\} , for all A,B\in\mathcal{P}(Z) , and the product with a nonnegative real number tA=\{ta:a\in A\} . We set A+\emptyset=\emptyset+A=\emptyset , for all A\in\mathcal{P}(Z) , t\emptyset=\emptyset , for all t\in\mathbb{R}\setminus\{0\} and 0\emptyset=\{0\} . We write A+\{z\}=A+z , A+\{-z\}=A-z . We set, for A\neq\emptyset or A\neq Z , (-1)A=-A=\{-a:a\in A\} , whereas -Z=\emptyset and -\emptyset=Z . Finally, we write A+(-B)=A-B . In KTZ2015 several order relations which are used in order to formulate corresponding solution concepts for a set-valued optimization problem are introduced, allowing more practical comparisons among values of the set-valued objective function. Let K\subset Z be a convex cone with \{0\}\varsubsetneq K\varsubsetneq Z . Then K induces in Z a quasiorder (reflexive and transitive binary relation) in the following way: x\leq_{K}y if and only if y-x\in K , for all x,y\in Z . Moreover, K also induces a quasiorder relation in \mathcal{P}(Z) as A\leq_{K}B if and only if B\subseteq A+K , see ( KTX1997 , Definition 2.1) , called lower set less order relation in ( KTZ2015 , Definition 2.6.9) . If we consider the subset of \mathcal{P}(Z) \mathcal{P}^{a}_{K}(Z):=\{{A\in\mathcal{P}(Z):A=A+K}\}, we will have that \leq_{K} induces a partial order (reflexive, antisymmetric and transitive binary relation) in \mathcal{P}^{a}_{K}(Z) , in fact, for A,B\in\mathcal{P}^{a}_{K}(Z) , A\leq_{K}B if and only if A\supseteq B . With the Minkowski sum and a little modification in the scalar product, 0A=K , for all A\in\mathcal{P}^{a}_{K}(Z) , we will have that \mathcal{P}^{a}_{K}(Z) is a partially ordered conlinear space (see H2005 for the definition and structural properties of conlinear spaces, also ( H2009 , Appendix) for a short introduction). Moreover, (\mathcal{P}^{a}_{K}(Z),\leq_{K}) is a complete lattice, where, for \mathcal{A}\subseteq\mathcal{P}^{a}_{K}(Z) , \sup\{\mathcal{A},\leq_{K}\}=\cap_{A\in\mathcal{A}}A and \inf\{\mathcal{A},\leq_{K}\}=\cup_{A\in\mathcal{A}}A . In the case \mathcal{A} is empty, we put \sup\{\mathcal{A},\leq_{K}\}=Z and \inf\{\mathcal{A},\leq_{K}\}=\emptyset . The following definitions are well-known. A set-valued function f:X\to\mathcal{P}(Z) is said to be proper if its effective domain \dom f:=\{x\in X:f(x)\neq\emptyset\} is nonempty and f(x)\neq Z , for all x\in\dom f . If f is not proper, we say that it is improper . It is said to be convex ( closed , resp.) if its graph \gph f:=\{{(x,z)\in X\times Z:z\in f(x)}\} is a convex (closed with respect to the product topology, resp.) set in X\times Z . We associate to any set-valued function f:X\to\mathcal{P}(Z) the set-valued function f_{K}:X\to\mathcal{P}^{a}_{K}(Z),f_{K}(x):=f(x)+K. (1) Definition 1 Let f:X\to\mathcal{P}(Z) be a set-valued function. A set-valued minorant of f is g:X\to\mathcal{P}^{a}_{K}(Z) verifying that, for all x\in X , g(x)\leq_{K}f(x) ,

Read the complete text