A theorem of the alternatives for the equation<i>Ax</i>+<i>B</i>|<i>x</i>| =<i>b</i> (Q4655433)

The main purpose of this work is to present a theorem of the alternatives for a nonlinear equation of the type \(Ax+B| x| =b\), where \(A,B\in \mathbb{R} ^{n\times n}\). Equations of this form arise quite naturally in solving systems of interval linear equations. The theorem says that given two matrices \(A,D\in \mathbb{R} ^{n\times n}\), \(D\geq 0\), then exactly one of the two alternatives holds: a) for each \(B\in \mathbb{R} ^{n\times n}\), with \(| B| \leq D\), where \(| B| =( | b_{ij}| ) \), and for each \( b\in \mathbb{R} ^{n}\) the above equation has a unique solution, and b) there exist \(\lambda \in [ 0,1] \) and a \(\pm 1\)-vector \(y\), such that the equation \( Ax+\lambda \operatorname{diag}( y) D| x| =0\) has a nontrivial solution. The alternative of (b) can also be reformulated in two equivalent ways. Finally, it is shown that for given rational data \(A\) and \(D\), the problem of determining which one of the alternatives (a), (b) holds is \(NP\)-hard.