A New Method for Large Scale Nonnegative Least Squares Problems

We present a new method for solving large scale nonnegative least squares problems. Firstly, nonnegative least squares problem was transformed into monotone linear complementarity problem. Then we apply potential-reduction interior point algorithm to monotone linear complementarity problem which is based on the Newton direction and centering direction. We show that this algorithm have the polynomial complexity. Numerical results are reported which demonstrate very good computational performance on nonnegative least squares problems.