“…However, since the matrix G is usually sparse, iterative methods are the usual choices for large-scale problems [8,9,16]. Taking into account that several authors reported that iterative methods based on Krylov subspaces, such as the conjugate gradient method for the normal equations and the LSQR method [10], performed better than stationary iterative methods such as SIRT and ART techniques [9,7,19], we focus on conjugate gradient methods in this work.…”