Modifications of the Oettli-Prager theorem with application to the eigenvalue problem
Given an \(n\times n\) interval matrix \([A]\) and an interval vector \([b]\) with \(n\) components the Oettli-Prager theorem characterizes the solution set \(S= \{x\in \mathbb{R}\mid x= Ax+ b\), \(A\in[A]\), \(b\in[b]\}\) by means of a vector inequality. Using the Fourier-Motzkin elimination process of linear programming one ends up with a characterization of \(S\) by linear inequalities which are equivalent to that of the Oettli-Prager theorem. This latter approach can be generalized to solution sets \(S_{\mathcal P}= \{x\in\mathbb{R}^n\mid Ax= b\), \(A\in[A]\), \(b\in [b]\), \(A\) with property \({\mathcal P}\}\subseteq S\), where \({\mathcal P}\) is some given property for \(A\in[A]\) such as symmetry, e.g. Therefore, it can be thought of a modification of the Oettli-Prager theorem. Applying both ideas to the equation \((A-\lambda I)x=0\) results in a characterization of the eigenpair sets \(E:= \{(x,\lambda)\in \mathbb{R}^{n+1}\mid Ax=\lambda x\), \(x\neq 0\), \(A\in[A]\}\), and \(E_{\text{sym}}:= \{(x,\lambda)\in \mathbb{R}^{n+1}\mid Ax=\lambda x\), \(x\neq 0\), \(A= A^T\in [A]\}\), respectively. In the paper several equivalences of the Oettli-Prager theorem are given, the Fourier-Motzkin elimination process is illustrated for symmetric matrices \(A\in [A]\) and the analogies for the algebraic eigenvalue problem are elaborated. An example for the latter problem concludes the paper. For the entire collection see [Zbl 0963.00018].
This page was built for publication: Modifications of the Oettli-Prager theorem with application to the eigenvalue problem
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2738476)