Locally controllable regularization of Abel integral equations via operator complementation

From MaRDI portal
Publication:4385203





The method of locally controllable regularization is described and analyzed for Abel's integral equation \[ (A\varphi)(x)\equiv {1\over\Gamma(\alpha)} \int^t_0 (x- t)^{\alpha- 1}\varphi(t) dt= f(x),\tag{1} \] where \(0\leq x\leq l<\infty\), \(\Omega= [0,l]\), \(0<\alpha\leq 1\). The perturbed data function is assumed to be given in \(L_q\), the true solution is assumed to lie in a Lipschitz space \(H^\lambda_p\), \(1\leq p\leq q\leq\infty\), \(\lambda>0\). Let \(\widetilde f\in L_q(\Omega)\) be a noize-perturbed function, which is not necessarily in the range of the linear operator \(A\). Let \(L(\Omega)\) be the set of all subsets \(G\subset\Omega\) with finite positive Lebesgue measure. A family of operators \(R_\delta\), \(\delta>0\) is called a ``locally controllable regularizing family for equation (1) on the set \(U\subset L_p(\Omega)\) if there exists a positive function \(\delta_{U,p,q}(\varepsilon, G)\) defined for \(\varepsilon>0\), \(G\in L(\Omega)\) and possessing the following property: For any \(\varphi_0\in U\) with \(\|\widetilde f- A\varphi_0\|_{L_q(\Omega)}< \varepsilon\) and for any \(G\in L(\Omega)\) it exists \(\delta= \delta_{U,p,q}(\varepsilon, G)\) such that \[ \| R_\delta\widetilde f- \varphi_0\|_{L_p(G)}\leq \gamma(U, \varepsilon, G,p,q)\to 0 \] as \(\varepsilon\to 0\). The error of the regularized approximate solution is measured in \(L_p\) norm corresponding to a subinterval of \(\Omega\). The approximation scheme is a finite difference imitation (of order determined by \(\lambda\)) to Abel's classical solution formula.











This page was built for publication: Locally controllable regularization of Abel integral equations via operator complementation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4385203)