On tensor product approximation of analytic functions (Q281571)

Let \(T>0\), \(\mathbf{a}\in\mathbb{R}_{+}^{d}\), and \(\mathcal{D}_{\mathbf{a}} (T) = \{ \mathbf{k}\in\mathbb{N}_{0}^{d} : \sum_{j=1}^{d} a_j k_j \leq T \}\). The paper contains matching upper and lower asymptotic bounds with explicit constants for sums of the form \(\sum_{ \mathbf{k}\in\mathbb{N}_{0}^{d} \setminus \mathcal{D}_{\mathbf{a}} (T) } \exp \left( - \sum_{j=1}^{d} a_j k_j \right) \), as \(T\rightarrow\infty\), which are useful in the error analysis of tensor product approximation of multivariate analytic functions. Applications are given to multivariate approximation by Taylor polynomials, as well as interpolation and integration at sparse tensor products of Leja points.