Single projection Kaczmarz extended algorithms (Q342879)

The extended Kaczmarz algorithm provides a tool for searching the least squares solution of a large inconsistent system of equations. It is known that the original Kaczmarz method does not converge to the least squares solution in the inconsistent case. The extended version of the Kaczmarz algorithm adds a correction step in the form of an orthogonal projection on the \(i\)-th hyperplane. When the projection indices are selected at random, the so-called randomized extended Kaczmarz algorithm converges in expectation to the minimal norm solution and convergence rates can be obtained. The presented paper introduces two deterministic control schemes, the maximal-residual and the almost cyclic control scheme and shows convergence to the least squares solution.