A hybrid algorithm for solving linear inequalities in a least squares sense (Q1014358)

New algorithms for solving the least deviation problem for inconsistent systems of linear inequalities \(Ax\geq b\), where \(A\) is a real \(m\times n\) matrix, \(x\in\mathbb{R}^n\), \(b\in\mathbb{R}^m\) is presented. The problem consists in minimization of the square of the Euclidean norm of \(y\in\mathbb{R}^m\) under the conditions \(Ax+ y\geq b\), \(y\geq 0\). This problem is appropriately reformulated, and the reformulated problem is solved by 'a new iterative algorithm -- so-called fixed matrix algorithm. In the next part of the paper, the fixed matrix algorithm is combined with the Newton's algorithm to obtain another algorithm -- the hybrid algorithm, which takes advantage of the both algorithms. Numerical experiments reported in the concluding part of the paper demonstrate the performance of the proposed algorithms.