Localized solutions to interval linear equations (Q1758400)

Given an \(m\times n\) interval matrix \({\mathbf A}=[\underline A,\overline A]\) and an interval vector \({\mathbf b}= [\underline b,\overline b]\) with \(m\) components, the authors consider the following particular solution sets: the \(L\)-localized solution set \(\Sigma_{LS}({\mathbf A},{\mathbf b})= \{x\in\mathbb{R}^n\mid{\mathbf A}\cdot x\leq_{st}{\mathbf b}\}\), the \(R\)-localized solution set \(\Sigma_{RS}({\mathbf A},{\mathbf b})= \{x\in\mathbb{R}^n\mid(-{\mathbf A}\cdot x)\leq_{st}(-{\mathbf b})\}\), and the localized solution set \(\Sigma_{ST}({\mathbf A},{\mathbf b})= \{x\in\mathbb{R}^n|\) for every J\(i\in\{1,\dots, m\}\) either \(({\mathbf A}\cdot x)_i\leq_{st}{\mathbf b}_i\) or \((-{\mathbf A}\cdot x)_i\leq_{st}(-{\mathbf b})_i\) holds\}, where the product \({\mathbf A}\cdot x\) is the usual one in interval analysis. The relation \({\mathbf a}\leq_{st}{\mathbf c}\) between two nonempty compact intervals \({\mathbf a}= [\underline a,\overline a]\), \({\mathbf c}= [\underline c,\overline c]\) is defined by \(\underline a\leq\underline c\leq\overline a\leq\overline c\); for vectors this relation has to hold componentwise. The authors illustrate and motivate these solution sets by examples. They prove two necessary and sufficient criteria for \(x\in\mathbb{R}^n\) to belong to \(\Sigma_{LS}({\mathbf A},{\mathbf b})\), to \(\Sigma_{RS}({\mathbf A},{\mathbf b})\), and to \(\Sigma_{ST}({\mathbf A},{\mathbf b})\), respectively, and they also state and prove a necessary one.