Solving large-scale constrained least-squares problems. (Q1406109): Difference between revisions

An algorithm for solving large scale regularized least-squares problems subject to quadratic inequality constraints is presented. The algorithm is based on recasting the considered problem into a parameterized eigenvalue problem. The main contribution is twofold. The first is related with the given comprehensive understanding of the problem, and the second is related with the development of different solutions of the problem. A convergence analysis for the considered algorithm is given. It is proved that this algorithm converges superlinearly.