Letter to the EditorPositive definite multi-kernels for scattered data interpolations
Introduction
In many areas of practical applications, we often face a problem of reconstructing an unknown function f from the scattered data. The scattered data consist of high-dimension data points and data values such that for all . The reconstruction is to find an estimate function s to approximate f. Generally, s is sought to interpolate the scattered data, that is, for all . It is well-known that the kernel-based approximation method is a fundamental approach of scattered data interpolations. The classical kernel-based approximation method is mainly dependent of the positive definite kernels. Here, we develop a concept of positive definite multi-kernels in Definition 3.1 which is a generalization of positive definite kernels. The positive definite multi-kernels can be also used to reconstruct f from the scattered data. For examples, we compare the interpolations by a positive definite kernel and a positive definite multi-kernel . The classical interpolant is composed of a kernel basis that is, where the coefficients are solved by a linear system The new interpolant is composed of another kernel basis that is, where the coefficients are solved by a multi-linear system In this article, we mainly discuss how to use a special class of strictly positive definite multi-kernel in Equation (3.3) to construct the kernel-based interpolant in Theorem 4.1 which shows that the related multi-linear system exists the unique solution. By the theorems of reproducing kernel Banach spaces in [5], we can obtain the advanced properties of including optimal recoveries in Theorem 4.3 and error analysis in Theorem 4.5, Theorem 4.6.
Section snippets
Positive definite tensors and multi-linear systems
In this section, we review the theory of positive definite tensors which will be used to define the positive definite multi-kernels in Section 3. For convenience of the readers, the notations and operations of tensors are defined as in the book [2]. We say a collection of all mth order nth dimensional real tensors, for example, , , and . For , we say a symmetric tensor if all entries of are invariant under any permutation of the indices. For
Positive definite multi-kernels and reproducing kernel Banach spaces
In this section, we develop a concept of positive definite multi-kernels. Let a domain . By [4, Definition 6.24], a symmetric kernel is said a positive definite kernel if, for all and all pairwise distinct points , the quadratic form for all . Let . Thus, is a symmetric positive definite matrix.
A multi-kernel of order is defined as for . We say a symmetric
Interpolations, optimal recoveries, and error analysis
In this section, we discuss how to construct the interpolant from the scattered data by the strictly positive definite multi-kernel in Equation (3.3). Suppose that the data compose of the pairwise distinct points and the values evaluated by some function , that is, Let . Let be a tensor defined in Equation (3.1) by the multi-kernel and the data points . Different
Acknowledgements
The author would like to acknowledge support for this project from the Natural Science Foundation of China (12071157 and 12026602), the Natural Science Foundation of Guangdong Province (2019A1515011995), and the Foundation of Department of Education of Guangdong Province (2020ZDZX3004).
References (6)
- et al.
Generalized tensor equations with leading structured tensors
Appl. Math. Comput.
(2019) - et al.
Solving multi-linear systems with M-tensors
J. Sci. Comput.
(2016) - et al.
Tensor Eigenvalues and Their Applications
(2019)