On iterative algorithms for linear least squares problems with bound constraints (Q756372)

This paper has been inspired by recent developments in trust region methods for solving nonlinear optimization problems with simple bounds. Two new projected gradient methods are described for solving the constrained least-squares problem: \(Min\{\| Ax-b\|:\ell \leq x\leq u\}\). Which of the two methods is preferable is problem dependent. However, their performance on a wide class of problems is claimed to be superior to the authors' improvement (for large scale problems) of an algorithm proposed by \textit{P. Lötstedt} [BIT 24, 206-224 (1984; Zbl 0546.65041)].