On the approximation of function by random splines of minimal defect (Q2739974)

Let \(\widetilde C=\widetilde C[0,2\pi]\) be the Banach space of \(2\pi\)-periodic continuous on \([0,2\pi]\) functions. Let us denote by \(\widetilde S_{2n,m}(\Delta_{2n})\) the space of \(2\pi\)-periodic splines of order \(m\) and defect 1 over homogeneous partition \(\Delta_{2n}:t_{i}=\pi i/n\), \(i=0,1,\ldots,2n\). Let \(\tau_{i}= \pi i/n-(1+(-1)^{m})\pi/4n\), \(i=1,2,\ldots,2n\) and let us suppose that in the points \(\tau_1,\tau_2,\ldots,\tau_{2n}\) we have to approximate the value of \(f(t)\): \(\widetilde y_{i}=f(\tau_{i})+\xi_{i}\), \(i=1,\ldots,2n\), where \(\xi_{i}\), \(i=1,2,\ldots,2n\) is a sequence of Gaussian independent random variables with parameters \(0,\sigma_2\). The author studies the estimate for approximation error \(\widetilde R(t)=f(t)-\sigma_{2n,m}(\widetilde y,t)\), where \(f(t)\in\widetilde C\) and \(\sigma_{2n,m}(\widetilde y,t)\) is the corresponding spline satisfying the equality \(\sigma_{2n,m}(\widetilde y,\tau_{i})=\widetilde y_{i}, i=1,2,\ldots,2n\).