A remark on the numerical solution of linear inverse problems with discrete data

The author presents a remark on a method for reducing the computational cost entailed in choosing, by the generalised cross-validation method, the regularisation parameter in the approximate solution of inverse problems.