Iterative methods for solving restricted linear equations (Q1126567)

Stationary iterative methods of the form \( x_{k+1} = x_k + \beta Z (b - Ax)\), \({\mathcal R}(Z) \subset T\), are investigated for computing some solution of the linear system of equations \(Ax=b\), \(A\in\mathbb{C}^{m\times n}\), such that the solution \(x\) is in a prescribed subspace \(T\subset\mathbb{C}^n\). Necessary and sufficient conditions for the convergence of the methods are proved. Three applications are discussed: (1) a rectangular inconsistent system (least squares problem), (2) a rectangular consistent system with \(T\cap{\mathcal N}(A) = \{ 0\}\), i.e.\ with unique solution, and (3) \(T={\mathcal R}(A^k)\) where \(k\) is the index of \(A\). For these applications possible choices for the matrix \(Z\) and scalar \(\beta\) are given that insure convergence.