GCV for Tikhonov regularization by partial SVD (Q1689313)

A new approach for the bounding of the Generalized Cross Validation (GCV) function is proposed in this paper. The developed approximation method is based on the observation that for many linear discrete ill-posed problems with a large matrix, it is quite inexpensive to determine the upper and lower bounds for the GCV function. The authors focus on the solution of linear discrete ill-posed problems with a large matrix \(A\), where the complete SVD computation is not feasible because of the size of \(A\). The identification of these bounds leads to the determination of a suitable value of the regularization parameter. The performance of the proposed method is demonstrated by the help of examples.