A Truncation Algorithm for Minimizing the Frobenius-Schatten Norm to Find a Sparse Matrix

Abstract

A problem of optimizing a matrix sparse in the joint Frobenius-Schatten norm is considered. The least rows are proposed to be truncated according to the lower bound to fight the ill-conditionality of the matrix. Truncation not only helps avoid incorrect termination of the algorithm but it also reduces the computational complexity. Convergence analysis ensures that a truncation algorithm finds an approximate solution to the problem. The numerical experiments show the advantage of the truncation method over the previous algorithm.