Discrete convex subdifferentials and optimality (Q6967159)

Given a set \(S\subset\mathbb{Z}^{n},\) one denotes by \(\mathrm{conv}_{\mathbb{R}^{n} }(S)\) its convex hull as a subset of \(\mathbb{R}^{n}.\) The set \(S\) is called convex in \(\mathbb{Z}^{n}\) if \((1-\lambda)x+\lambda y\in S\) for all \(x,y\) in \(\mathrm{conv}_{\mathbb{R}^{n}}(S)\cap\mathbb{Z}^{n}\) and \(\lambda\in\lbrack0,1]\) such that \((1-\lambda)x+\lambda y\in\) \(\mathbb{Z}^{n}\). An extended real-valued function \(f:\mathbb{Z}^{n}\rightarrow\mathbb{R\cup\{\infty\}}\) is called convex if its epigraph \(\left\{ (x,\alpha)\in\mathbb{Z}^{n}\times \mathbb{R}:f(x)\leq\alpha\right\} \) is a convex set in \(\mathbb{Z}^{n} \times\mathbb{R}\). Based on previous work done by \textit{B.~Mordukhovich} and \textit{N.~M. Nam} [An easy path to convex analysis and applications. 2nd edition. Cham: Springer (2023; Zbl 1515.49004)], the authors have extended the notions of normal cone, subgradient and optimality conditions from the setting of \(\mathbb{R}^{n}\) to \(\mathbb{Z}^{n}\). Various properties and calculus rules of normal cones, subdifferentials and singular subdifferentials are established and the differences with existing analogous rules in the real domain \(\mathbb{R}^{n}\) are justified through counter-examples. The paper also includes optimality conditions for a constrained optimization problem.