Quadrature formulas for classes with constraints on the Fourier coefficients (Q1814470)

Let \(\alpha_ 1,\dots,\alpha_ s\in \mathbb{N}\), \(\kappa\in [2;\infty]\), \(M>0\), and let \(\alpha=(\alpha_ 1,\dots,\alpha_ s)\). We define \(E^ \kappa_ \alpha(M)\) as the class of continuous functions \(f\) on \(\mathbb{R}^ s\) possessing period 1 with respect to each of the variables whose Fourier coefficients satisfy the condition \(\Bigl(\sum_{m\in \mathbb{Z}^ s}|\widehat f(m)\cdot\overline m^ \alpha|^ \kappa\Bigr)^{1/\kappa}\leq M\), where \(\overline m^ \alpha= \overline m^{\alpha_ s}_ s\) and \(\overline m_ j=\max(1,| m_ j|)\). In this way we introduce a new classification scale of periodic functions, depending on the rate and character of the decrease of their Fourier coefficients \(\dots\) We obtain a formula expressing the error of the approximate integration of \(E^ \kappa_ \alpha(M)\) on a net in terms of the values of the trigonometric sum of this net. For \(\kappa\geq 2\) we determine the correct order of the error of the approximate integration of the classes \(E^ \kappa_ \alpha(M)\) on nets with \(N\) nodes and we indicate the nets (variants of K. Frolov's algebraic net) for which this order is attained. The question of the correct order for \(\kappa\in [1;2)\) remains open.