English

Comparison of Two Search Criteria for Lattice-based Kernel Approximation

Numerical Analysis 2023-04-05 v1 Numerical Analysis

Abstract

The kernel interpolant in a reproducing kernel Hilbert space is optimal in the worst-case sense among all approximations of a function using the same set of function values. In this paper, we compare two search criteria to construct lattice point sets for use in lattice-based kernel approximation. The first candidate, \calPn\calP_n^*, is based on the power function that appears in machine learning literature. The second, \calSn\calS_n^*, is a search criterion used for generating lattices for approximation using truncated Fourier series. We find that the empirical difference in error between the lattices constructed using \calPn\calP_n^* and \calSn\calS_n^* is marginal. The criterion \calSn\calS_n^* is preferred as it is computationally more efficient and has a proven error bound.

Keywords

Cite

@article{arxiv.2304.01685,
  title  = {Comparison of Two Search Criteria for Lattice-based Kernel Approximation},
  author = {Frances Y. Kuo and Weiwen Mo and Dirk Nuyens and Ian H. Sloan and Abirami Srikumar},
  journal= {arXiv preprint arXiv:2304.01685},
  year   = {2023}
}

Comments

16 pages, 3 figures

R2 v1 2026-06-28T09:48:45.642Z