A mixed Newton-Tikhonov method for nonlinear ill-posed problems (Q735296)

In the last years, the theory and methods to solve nonlinear ill-posed problems has been developed extensively. A nonlinear ill-posed problem is a nonlinear operator equation with a nonlinear Fréchet differentiable mapping between two Hilbert spaces, in the left-hand side of the equation, and it is called ill-posed, in the sense that the solution does not depend continuously on the right-hand side data. In the realistic problems, the right-hand side of the equation is obtained by measurement and not accurate, so the equation has a perturbed right-hand side. In this paper, the authors apply the Tikhonov regularization method (which consists in finding the minimum of a certain functional) to the algorithm of the simplified Newton method for a well-posed problem. A mixed Newton-Tikhonov method to solve ill-posed problems is proposed. The main results of this paper are the convergence and the stability analysis of the mixed Newton-Tikhonov iteration, under noise-free data of the right-hand side of the operator equation, and also, the convergence analysis under noisy data of the right-hand side of the equation. Numerical experiments are made to solve an inverse heat conduction ill-posed problem by the mixed Newton-Tikhonov method proposed in this paper. This example shows that the new proposed method has obvious advantages over the classical Newton method in terms of computational costs.