1: \begin{abstract}

2: In this work we devise a new version of the Lawson-Hanson with Deviation Maximization (LHDM) algorithm, a block method for the solution of NonNegative Least Squares (NNLS). 

3: %This algorithm allows to exploit BLAS-3 operations, leading to higher performances \#F: per\'o in Matlab non riusciamo a vederlo sperimentalmente. 

4: It is shown that the new algorithm here presented convergences in a finite number of steps under suitable conditions. An extensive campaign of experiments is performed in order to evaluate the performance gain with respect to the standard Lawson-Hanson algorithm. We also explore the sparse recovery ability of LHDM, comparing it against several $\ell_1$-minimization solvers in terms of solution quality and time-to-solution on a large set of dense instances.

5: %LO SARA' NEL PAPER (NON NEL PREPRINT) An implementation of the proposed algorithm is available on a public repository.

6: \end{abstract}

7: