Regularization with Differential Operators: An Iterative Approach (Q4029161)

The author introduces iterative counterparts to the regularization method of Tikhonov-Philips with a general smoothing operator \(L\) to solve an ill-posed or ill-conditioned linear equation \(Kx = b\), that is, choose \(K\) such that \(\| b-Kx\|^ 2 + \alpha\| Lx\|^ 2 \to \min\). An explicit formula for the regularized approximation \(x_ \alpha\) is determined by (A) \(x_ \alpha = x_ 0 + L^ -r_ \alpha(L^{- *}K^*KL^ -)L^{-*}K^*b\), where \(L^ -\) is the generalized inverse of \(L\). Depending the choice of the function \(r_ \alpha\), the formula (A) can become the Tikhonov-Phillips method, the truncated singular value method, the semi-iterative method, or the preconditioned conjugate gradient method. Finally, the author presents a numerical example to illustrate the effectiveness of the regularization methods.