Optimal recovery of linear functionals of Peano type through data on manifolds (Q1904188): Difference between revisions

Let \(D\subset \mathbb{R}^n\) \((n\geq 2)\) be a bounded domain for which there exists a sequence \(D_1, \dots, D_{N-1}\) of domains satisfying \[ D_j\subset \overline {D}_j \subset D_{j+1} \qquad j=0, \dots, N-1, \] where \(D_0= \emptyset\) and \(D_N= D\). It is assumed that each of the sets \(\Gamma_j= \partial D_j\) \((j=1, \dots, N)\) is a \(C^\infty\) manifold. The author deals with the problem of optimal recovery of a linear functional \(J\), defined on the Sobolev space \(H^{4q} (D)\) by means of the sum of finitely many integrals of the type \[ \int_S f(x) m(x) d\sigma (x) \qquad (f\in H^{4q} (D)), \] where \(S\subset D\) is an \(r\)-dimensional compact manifold and \(m\in C(S)\). Such a functional is said to be of simple Peano type. For recovering \(J\), there are employed linear functionals that are defined on \(H^{4q} (D)\) by using the information \[ T(f)= (\pi_1 (f), \dots, \pi_N(f)), \] where \(\pi_j(f)\) \((j=1, \dots, N)\) is the trace of \(f\in H^{4q} (D)\) on the manifold \(\Gamma_j\). The procedure for finding a solution of the considered problem is similar to the method applied in the univariate case, but instead of natural splines it needs the polysplines previously introduced by the author [C. R. Acad. Bulg. Sci. 44, No. 7, 9-11 (1991; Zbl 0772.41014); ibid. 44, No. 8, 13-16 (1991; Zbl 0777.41014)].