One approach to the comparison of error bounds at a point and on a set in the solution of ill-posed problems (Q1800525)

In a special case the error estimates at a point and on a set in the case of an ill-posed problem are compared. Approximate solutions are constructed by a regularization method. Under consideration is the problem of calculation of the value \(u_{0}=Tf\), \(Tf(\xi)=t(\xi)f(\xi)\), \(T:L_{2}({\mathbb R})\to L_{2}({\mathbb R})\) with \(D(T)=\{f\in L_{2}({\mathbb R}):\;Tf\in L_{2}({\mathbb R})\}\), for a given point \(u_{0}\) and on a particular set \(u_{0}\in Z\subset L_{2}({\mathbb R})\). Here \(t(\xi)\) is a continuous function satisfying certain growth conditions at infinity. Regularized solutions to the problem are defined as \(u_{\alpha,\delta}(\xi)=Tf_{\delta}(\xi)\) for \(|\xi|\leq \alpha\) and \(u_{\alpha,\delta}=0\) for \(|\xi|>\alpha\), where \(f_{\delta}\) is an approximation of \(f\). The error estimate at a point \(u_{0}\) is defined as \( \Delta_{\delta}(u_{0})=\sup\{ \|Tf_{\delta}-Tf\|:\;f_{\delta}\in L_{2}({\mathbb R}), \|f_{\delta}-f\|<\delta\}. \) The error estimate on the set \(Z\) is defined similarly. It is demonstrated that the error estimate at a point is an infinitely small quantity in comparison with that on a set.