Regularization of integral equations in spaces of distributions (Q1429023): Difference between revisions
From MaRDI portal
Removed claims |
Changed an Item |
||
Property / author | |||
Property / author: Ricardo Estrada / rank | |||
Normal rank | |||
Property / reviewed by | |||
Property / reviewed by: Alexei Yu. Karlovich / rank | |||
Normal rank |
Revision as of 15:33, 11 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Regularization of integral equations in spaces of distributions |
scientific article |
Statements
Regularization of integral equations in spaces of distributions (English)
0 references
29 March 2004
0 references
For \(-\infty\leq a<b\leq+\infty\), consider four spaces of smooth functions: (1) Let \({\mathcal D}(a,b)\) be the space of smooth functions that vanish outside of \((a,b)\); (2) let \({\mathcal E}(a,b)\) be the space of smooth functions without any restriction on their supports; (3) let \({\mathcal E}[a,b]\) be the set of smooth functions on \([a,b]\), where being smooth at the endpoints means that the one-sided limits of the derivatives \(\varphi^{(j)}(a+0)\) and \(\varphi^{(j)}(b-0)\) exist for \(j=0,1,2,\dots\); and (4) let \({\mathcal S}(a,b)\) be the subset of \({\mathcal E}[a,b]\) formed by those functions for which \(\varphi^{(j)}(a+0)=\varphi^{(j)}(b-0)=0\) for each \(j=1,2,\dots\). For \(j,k=1,\dots,4\), let \({\mathcal D}_{jk}(a,b)\) denote the space of smooth functions on \((a,b)\) that satisfy condition \((j)\) at \(a\) and condition \((k)\) at \(b\). All above spaces of smooth functions can be equipped with natural topologies and one can consider corresponding spaces of distributions. Let \({\mathcal X}\) be a topological vector space and let \(T:{\mathcal X}\to {\mathcal D}_{3i}'(a,b)\) be a continuous operator. A function \(\psi\) defined on \((a,b)\) is called a multiplicative regularizator of \(T\) at \(a\) if \(\psi(x)>0\) for \(x\in(a,b)\) and there exists a continuous operator \(\widetilde{T}:{\mathcal X}\to{\mathcal D}_{4i}'[a,b)\) such that \(\pi(\widetilde{T}f)=\psi Tf\) for \(f\in{\mathcal X}\), where \(\pi:{\mathcal D}_{4i}'[a,b)\to{\mathcal D}_{3i}'(a,b)\) is the canonical projection. A function \(\psi\) is said to be a proper multiplicative regularizator if there is \(c\in(a,b)\) such that the image of \(\pi_c\widetilde{T}\) has finite codimension in \({\mathcal D}_{43}'[a,c)\), where \(\pi_c:{\mathcal D}_{4i}'[a,b)\to {\mathcal D}_{43}'[a,c)\) is the canonical projection. The author proves necessary conditions for a function \(\psi\) to be a multiplicative regularizator and a proper multiplicative regularizator of the projection \(\pi:{\mathcal E}'[a,b]\to{\mathcal D}_{34}'(a,b]\) (Theorem 2). He also proves the following ``uniqueness'' theorem. Theorem 3. Let \(T:{\mathcal E}'[a,b]\to{\mathcal D}_{34}'(a,b]\) be a continuous operator. Suppose \(\psi_1\) and \(\psi_2\) are two proper multiplicative regularizators of \(T\) with associated operators \(\widetilde{T}_1,\widetilde{T}_2: {\mathcal E}'[a,b]\to{\mathcal E}'[a,b]\). If \(\widetilde{T}_1\) and \(\widetilde{T}_2\) are both Fredholm operators, then the limit \(\lim\limits_{x\to a}\log[\psi_1(x)/\psi_2(x)]/\log(x-a)=k\) exists and \(k\in {\mathbb N}\). Finally, the author considers applications of the above results to the Hilbert transform, the Carleman operator, the operator of fractional integration, and the generalized Abel operator.
0 references
integral equation
0 references
distribution
0 references
Hilbert transform
0 references
Carleman operator
0 references
operator of fractional integration
0 references
generalized Abel operator
0 references
regularization
0 references