MATLAB routines for estimating the diagonal of the resolution matrix and the GCV function for large discrete linear inverse problems

This MATLAB code includes functions for computing the generalized cross validation (GCV) function and the diagonal of the resolution matrix for large scale discrete linear inverse problems that have been regularized by TIkhonov regularization. LSQR is used to solve the least squares problems. THe stochastic method of Bekas et al. is used to estimate the matrix diagonals. The method is described in a companion paper.

The code is available in

Return to the Brian Borchers Homepage