Applications of the Modified Discrepancy Principle to Tikhonov Regularization of Nonlinear Ill-Posed Problems (Q4243574)

The author considers the finite-dimensional approximations of Tikhonov regularization for nonlinear ill-posed problems with approximately given right-hand sides. The article is organized as follows. In section 2 a rule is proposed for Tikhonov regularization of nonlinear ill-posed problems in finite-dimensional settings to choose the regularization parameter, and the justification is given under appropriate conditions. Section 3 contributes to the analyses of the convergence and rates of convergence. The uniqueness of the Tikhonov regularized solution, which is crucial for these analyses, is considered in section 4. Finally, two concrete examples from the parameter estimations of partial differential equations are given in section 5 to illustrate the applicability of the results. The first example treats with the problem of determining the source term \(g\) in the equation \[ -\Delta u+ g\cdot\nabla u+ K(u)= g\quad\text{with }u|_{\partial\Omega}= 0 \] from the noisy data \(u_\delta\) of the state \(u_0\), which corresponds to the sought source term \(g_0\). \(\Omega\) is a bounded domain in \(\mathbb{R}^n\), \(n\leq 3\) with the cone property, and the nonlinear function \(K\) has the property that \(K\in W^{2,\infty}(\mathbb{R})\), \(K(0)= 0\), and \(K'(x)\geq 0\) for all \(x\in\mathbb{R}\) and the vector valued function \(q\) satisfies \(q\in W^{2,\infty}(\Omega,\mathbb{R}^n)\) with \(\nabla\cdot q=0\) on \(\Omega\). The second example treats with the problem of estimating the diffusion coefficient \(a\) in two point boundary value problem \[ -(au_x)_x= f\quad\text{in }(0,1)\quad\text{and }u(0)= u(1) \] with \(f\in L^2[0,1]\) from the noisy data \(u_\delta\) of the state \(u_0\), \(\|u_\delta- u_0\|_{L^2}\leq \delta\). This study extends the results, which were developed by \textit{R. Plato} and \textit{C. M. Vainikko} [Numer. Math. 57, No. 1, 63-79 (1990; Zbl 0675.65053)] for solving linear ill-posed equations, to nonlinear problems.